Patent Application
20220293274
The present invention relates to a method of predicting whether a kidney transplant recipient is at risk of having allograft loss.
End-stage renal disease is estimated to affect 7.4 million persons worldwide.(1, 2) According to data from the World Health Organisation, more than 1,500,000 live with transplanted kidneys, and 80,000 new kidneys are transplanted each year.(3) Despite the considerable advances in short-term outcomes, kidney transplant recipients continue to suffer from late allograft failure, and little improvement has been made over the past 15 years.(4, 5) While the failure of a kidney allograft represents nowadays an important cause of end stage renal disease, it contrasts with the absence of available robust and widely validated prognostication systems for the risk of allograft failure in individual patients.(6) Accurately predicting which patients are at a high risk of allograft loss would help to stratify patients into clinically meaningful risk groups, which may help guide patient monitoring. Moreover, regulatory agencies and medical societies have highlighted the need for an early and robust surrogate endpoint in transplantation that adequately predicts long-term allograft failure.(7) An enhanced ability to predict allograft outcomes would not only inform daily clinical care, patient counselling and therapeutic decisions but also facilitate the performance of clinical trials, which generally lack statistical power because of the low event rates during the first year after transplantation.(8)
Taken individually, parameters such as estimated glomerular filtration rate (eGFR),(9, 10) proteinuria,(11) histology,(12) or human leukocyte antigen (HLA) antibody profiles,(13) fail to provide sufficient predictive accuracy. Previous efforts at developing prognostic systems in nephrology based on various combinations of parameters have been hampered by small sample sizes, the absence of proper validation, limited phenotypic details from registries, the absence of systematic immune response monitoring, and the failure to include key prognostic factors that affect allograft outcome (e.g., donor-derived factors, polyoma virus-associated nephropathy, disease recurrence).(14-16) Finally, no scoring system has been evaluated in large cohorts from different countries with different transplant practices, allocation systems and practice patterns, thereby limiting their exportability, which is an important consideration for health authorities to accept a scoring system as a surrogate endpoint.(17)
As defined by the claims, the present invention relates to a method of predicting whether a kidney transplant recipient is at risk of having allograft loss.
Organ transplantation is currently recognised as the treatment of choice for patients with end-stage renal disease (ESRD), which is an underestimated but increasing burden worldwide. Indeed, chronic kidney disease (CKD) affects 850 million individuals worldwide (in comparison, diabetes, cancer, and HIV/AIDS affect 422, 42, and 37 million individuals worldwide, respectively). Despite the progress made in immunosuppressive regimens, thousands of allografts are failing every year, with immediate consequences for the patients in terms of mortality, morbidity and cost for the society. Recently, it has been shown that the loss of a kidney allograft represents nowadays an important cause of ESRD. Therefore, the possibility to identify accurately patients who are the most likely to lose their renal allograft carries major clinical implications.
Despite the pressing need for improving patients risk stratification raised by transplant societies (e.g., the European Society of Organ Transplantation, the American Society for Transplantation and the American Society of Transplant Surgeons), regulatory agencies (e.g., the European Medicine Agency and the U.S. Food & Drug Administration), no risk-stratification system exists that adequately predicts transplant patients' individual risk of allograft loss. This currently represents a limitation for improving patient management, as well as for defining early surrogate end points for clinical trials and development of pharmaceutical agents.
The inventors now report the development and validation of an integrative risk prediction score to predict kidney allograft survival of individual patients (NCT03474003). The iBox risk prediction score is the first integrative system validated in several independent populations from Europe & North America as well as across 3 clinical trials (NCT01079143, EudraCT2007-003213-13, NCT01873157) covering distinct clinical scenarios. In particular, the advantages brought by the iBox risk prediction score are i) improved discrimination performance by combining traditional prognostic factors with mechanistically informed parameters, ii) outperformance when compared with currently existing scoring systems, iii) generalisability when assessed in geographically distinct cohorts from Europe and North America, iv) transportability at different times of evaluation post-transplant, v) performance in a variety of clinical scenarios including clinical trials and vi) readily accessible to clinicians and patients by an online tool for patient risk calculation.
The first object of the present invention relates to a method, preferably an in vitro method, of predicting whether a kidney transplant recipient is at risk of having allograft loss comprising the steps of:
a) assessing for said recipient a plurality of parameters, said parameters being:
b) implementing an algorithm on data comprising or consisting of the parameters assessed at step a) as to obtain an algorithm output, the implementing step being computer-implemented; and
c) determining the risk of allograft loss, in particular at any time after transplantation, from the algorithm output obtained at step b).
As used herein, the term “recipient” refers to any subject, in particular a human subject, that receives an organ and/or tissue transplant or graft obtained from a donor. The term “donor” as used herein refers to the subject that provides the organ and/or tissue transplant or graft to be transplanted into the recipient. As used herein, the term “kidney transplant recipient” refers to an individual that has undergone kidney transplantation.
As used herein, the term “allograft loss” refers to loss of function in a transplanted organ. In kidney transplant recipients, graft loss often means return to dialysis.
As used herein, the term “risk” in the context of the present invention, relates to the probability that an event will occur over a specific time period, as in the conversion to allograft loss, and can mean a subject's “absolute” risk or “relative” risk. Absolute risk can be measured with reference to either actual observation post-measurement for the relevant time cohort, or with reference to index values developed from statistically valid historical cohorts that have been followed for the relevant time period. Relative risk refers to the ratio of absolute risks of a subject compared either to the absolute risks of low risk cohorts or an average population risk, which can vary by how clinical risk factors are assessed. Odds ratios, the proportion of positive events to negative events for a given test result, are also commonly used (odds are according to the formula p/(l−p) where p is the probability of event and (l−p) is the probability of no event) to no conversion. Accordingly, the expression “predicting whether a kidney transplant recipient is at risk of having allograft loss” in the context of the present invention encompasses making a prediction of the probability, odds, or likelihood that allograft loss may occur. The methods of the present invention may be used to make continuous or categorical measurements of the risk of conversion to allograft loss. -The method of the present invention is particularly suitable to predict the risk of allograft loss at 3, 5, and 7 years from the date of prediction.
As used herein, the term “parameter” refers to any characteristic tested when carrying out the method according to the invention. As used herein, the term “parameter value” refers to a value (a number for instance) associated to a parameter.
As used herein, the term “time from transplant to risk evaluation” or “time of posttransplant risk evaluation” refers to the time that is comprised between the 1 month posttransplantation and the day of the risk evaluation. Typically, the time from transplant risk evaluation is comprised between 1 month and 120 months.
According to the present invention, allograft functional parameters comprise or consist of estimated or measured glomerular filtration rate and proteinuria.
As used herein, the term “glomerular filtration rate” or “GFR” refers to the volume of fluid filtered from the renal (kidney) glomerular capillaries into the Bowman's capsule per unit time. GFR is used to assess renal function in a subject. As used herein, the term “estimated GFR” or “eGFR” refers to an estimate of the Glomerular Filtration Rate or GFR, calculated using the Modification of Diet in Renal Disease (MDRD) equation developed by the Modification of Diet in Renal Disease Study Group described in Levey A S, Bosch J P, Lewis J B, Greene T, Rogers N, Roth D, “A more accurate method to estimate glomerular filtration rate from serum creatinine: a new prediction equation. Modification of Diet in Renal Disease Study Group” Ann. Intern. Med. 130 (6): 461-70 (1999), the contents of which are herein incorporation by reference. It also refers to measured GFR clearance using exogenous filtration markers that are eliminated exclusively by glomerular filtration, including inulin, iohexol, iothalamate, technetium 99 m diethylenetriamine pentaacetic acid (99mTc-DTPA) and chromium 51-ethylenediamine tetra-acetic acid (51Cr-EDTA). Typically, the unit of measurement for GFR (measure or estimated) is mL/min/1,73 m2. Typically, the eGFR is comprised between 0 and 120 mL/min/1,73 m2.
As used herein, the term “proteinuria” refers to a condition in which excess protein is present in the urine of a subject. In human subjects, proteinuria is often diagnosed by urinalysis. Clinically, different approaches are used to measure proteinuria, including:
According to the present invention, allograft histological parameters comprise or consist of interstitial fibrosis and tubular atrophy (IFTA), glomerulitis and peritubular capillaritis, interstitial inflammation and tubulitis, and transplant glomerulopathy but also intimal arteritis, C4d, vascular fibrous intimal thickening, mesangial matrix expansion, arteriolar hyalinosis, hyaline arteriolar thickening, total inflammation, inflammation in the area of IFTA, tubulitis in the atrophic tubules.
According to the present invention, the allograft histological parameters are assessed according to the Banff Classification that is an international consensus classification for the reporting of biopsies from solid organ transplants. Banff Lesion Scores indeed assess the presence and the degree of histopathological changes in the different compartments of renal transplant biopsies, focusing primarily but not exclusively on the diagnostic features seen in rejection.
In particular, one of the allograft histological parameter is the interstitial fibrosis/tubular atrophy (IFTA) that is evaluated by Banff Lesion Score IFTA. This score evaluates the extent of inflammation in scarred cortex. The score is assessed as follows:
Another allograft histological parameter is microcirculation inflammation (corresponding to the combination glomerulitis and peritubular capillaritis) that results from the addition of Banff Lesion Score g (score for glomerulitis)+Banff Lesion Score ptc (score for peritubular capillaritis).
Banff Lesion Score g evaluates the degree of inflammation within glomeruli. Glomerulitis is a form of microvascular inflammation and is a feature of activity and antibody interaction with tissue in antibody-mediated rejection. The score is assessed as follows:
Banff Lesion Score ptc evaluates the degree of inflammation within peritubular capillaries (PTCs). Together with glomerulitis, peritubular capillaritis constitutes microvascular inflammation as a feature of active antibody-mediated rejection or chronic active antibody-mediated rejection. The score is assessed as follows:
Another allograft histological parameter is the interstitial inflammation and tubulitis that results from the addition of Banff Lesion Score i (score for interstitial inflammation)+Banff Lesion Score t (score for tubulitis).
Banff Lesion Score i evaluates the degree of inflammation in nonscarred areas of cortex (“interstitial Inflammation”), which is often a marker of acute T cell—mediated rejection. The score is assessed as follows:
Banff Lesion Score t evaluates the degree of inflammation within the epithelium of the cortical tubules (“tubulitis”). The presence of mononuclear cells in the basolateral aspect of the renal tubule epithelium is one of the defining lesion of acute T cell-mediated rejection in kidney transplants. The score is assessed as follows:
Another allograft histological parameter is the transplant glomerulopathy (cg) that is evaluated by Banff cg Score. The score is based on the presence and extent of glomerular basement membrane (GBM) double contours or multilamination in the most severely affected glomerulus. The score is assessed as follows:
In a particular embodiment, said allograft histological parameters further include inflammation in areas of IFTA (i-IFTA), C4d staining (C4d), vascular fibrosis intimal thickening (cv), arteriolar hyalinosis (ah), mesangial matrix expansion (mm), tubulitis in atrophic tubules (tIFTA).
C4d staining (C4d) is evaluated by Banff C4d score. The score is based on the extent of staining for C4d on endothelial cells of PTCs and medullary vasa recta by IF on snap frozen sections of fresh tissue or IHC on formalin-fixated and paraffin-embedded tissue. The score is assessed as follows:
Vascular fibrosis intimal thickening (cv) is evaluated by Banff cv score. It reflects the extent of arterial intimal thickening in the most severely affected artery. The score is assessed as follows:
Mesangial matrix expansion (mm) is evaluated by Banff mm score. It evaluates the percentage of glomeruli with “moderate mesangial matrix expansion” in relation to all nonsclerosed glomeruli. The score is assessed as follows:
Arteriolar hyalinosis (ah) is evaluated by Banff ah score. It evaluates the extent of arteriolar hyalinosis. The score is assessed as follows:
Inflammation in areas of IFTA (i-IFTA) is evaluated by Banff i-IFTA score. It evaluates the extent of inflammation in scarred cortex. The score is assessed as follows:
Tubulitis in atrophic tubules (t-IFTA) is evaluated by Banff t-IFTA score. It evaluates tubulitis in atrophic tubules. The score is assessed as follows:
According to the present invention, the allograft histological parameters can also be assessed according to the Banff Classification with diagnosis labels:
A further parameter assessed in the method of the invention is the recipient immunological profile that comprises or consists of the presence and level of the immunodominant circulating anti-HLA donor-specific antibodies (DSA). As used herein, the term “anti-HLA DSA” has its general meaning in the art and refers to the donor-specific anti-HLA antibodies present in the subject. There are several sensitive tests known by the skilled man to determine the level of de novo donor-specific anti-HLA antibodies. For instance, an example of a test to determine the level of de novo donor-specific anti-HLA antibodies comprises: screening of antibodies to HLA-A, HLA-B, HLA-C, HLA-DP, HLA-DQ and HLA-DR gene products using Luminex® solid-phase assay (one lambda Labscreen assay) on serum samples. Typically, the level if expressed as mean-fluorescence intensity (“MFI”) is comprised between 0 and 10 000.
In a particular embodiment, no additional parameter is assessed and/or used in the algorithm, in the prediction method of the invention.
As used herein, the term “algorithm” is any mathematical equation, algorithmic, analytical or programmed process, or statistical technique that takes one or more continuous parameters and calculates an output value, sometimes referred to as an “index” or “index value.” Non-limiting examples of algorithms include sums, ratios, and regression operators, such as coefficients or exponents, biomarker value transformations and normalizations (including, without limitation, those normalization schemes based on clinical parameters, such as gender, age, or ethnicity), rules and guidelines, statistical classification models, and neural networks trained on historical populations. Of particular use in combining parameters are linear and non-linear equations and statistical classification analyses to determine the relationship between levels of said parameters and the risk of allograft loss. Of particular interest are structural and syntactic statistical classification algorithms, and methods of risk index construction, utilizing pattern recognition features, including established techniques such as cross-correlation, Principal Components Analysis (PCA), factor rotation, Logistic Regression (LogReg), Linear Discriminant Analysis (LDA), Eigengene Linear Discriminant Analysis (ELDA), Support Vector Machines (SVM), Random Forest (RF), Recursive Partitioning Tree (RPART), as well as other related decision tree classification techniques, Shrunken Centroids (SC), StepAIC, Kth-Nearest Neighbor, Boosting, Decision Trees, Neural Networks, Bayesian Networks, Support Vector Machines, and Hidden Markov Models, among others. Other techniques may be used in survival and time to event hazard analysis, including Cox, Weibull, Kaplan-Meier and Greenwood models well known to those of skill in the art.
In some embodiments, the method of the present invention comprises the use of a machine learning algorithm. The machine learning algorithm may comprise a supervised learning algorithm. Examples of supervised learning algorithms may include Average One-Dependence Estimators (AODE), Artificial neural network (e.g., Backpropagation), Bayesian statistics (e.g., Naive Bayes classifier, Bayesian network, Bayesian knowledge base), Case-based reasoning, Decision trees, Inductive logic programming, Gaussian process regression, Group method of data handling (GMDH), Learning Automata, Learning Vector Quantization, Minimum message length (decision trees, decision graphs, etc.), Lazy learning, Instance-based learning Nearest Neighbor Algorithm, Analogical modeling, Probably approximately correct learning (PAC) learning, Ripple down rules, a knowledge acquisition methodology, Symbolic machine learning algorithms, Subsymbolic machine learning algorithms, Support vector machines, Random Forests, Ensembles of classifiers, Bootstrap aggregating (bagging), and Boosting. Supervised learning may comprise ordinal classification such as regression analysis and Information fuzzy networks (IFN). Alternatively, supervised learning methods may comprise statistical classification, such as AODE, Linear classifiers (e.g., Fisher's linear discriminant, Logistic regression, Naive Bayes classifier, Perceptron, and Support vector machine), quadratic classifiers, k-nearest neighbor, Boosting, Decision trees (e.g., C4.5, Random forests), Bayesian networks, and Hidden Markov models. The machine learning algorithms may also comprise an unsupervised learning algorithm. Examples of unsupervised learning algorithms may include artificial neural network, Data clustering, Expectation-maximization algorithm, Self-organizing map, Radial basis function network, Vector Quantization, Generative topographic map, Information bottleneck method, and IBSEAD. Unsupervised learning may also comprise association rule learning algorithms such as Apriori algorithm, Eclat algorithm and FP-growth algorithm. Hierarchical clustering, such as Single-linkage clustering and Conceptual clustering, may also be used. Alternatively, unsupervised learning may comprise partitional clustering such as K-means algorithm and Fuzzy clustering. In some instances, the machine learning algorithms comprise a reinforcement learning algorithm Examples of reinforcement learning algorithms include, but are not limited to, temporal difference learning, Q-learning and Learning Automata. Alternatively, the machine learning algorithm may comprise Data Pre-processing.
In a particular embodiment, the algorithm is a classifier.
In some embodiments, the output obtained by the algorithm at step b) is a score. As used herein, the term “score” refers to a piece of information, usually a number that conveys the result of the subject on a test. A risk scoring system separates a patient population into different risk groups; herein the process of risk stratification typically classifies the patients into very high-risk, high-risk, intermediate-risk and low-risk groups.
In some embodiments, the score corresponds to the score depicted in EXAMPLE 2. In particular embodiments, the score is determined using the algorithm which consists in applying the following formula:
More generally, the score is a weighted sum of one or several of a function applied to a specific assessed parameter. The function is linear or logarithmic in the previous case.
In the current case, the weight are the Cox-model beta coefficients but any model providing with weight for the assessed parameters can be used.
Notably, the assessed parameters can be learned by an artificial intelligence technique.
In a preferred embodiment, the algorithm comprises using a survival and time to event hazard model, such as the Cox model, wherein the relationship (in the sense of the Cox model) between predictors and allograft loss is approximated as either linear or polynomial.
For predictors with a polynomial relationship with the allograft loss, a fractional polynomial method was applied to obtain a relationship. In the present case, these predictors are the continuous predictors.
The quality of such regression is challenged by using several techniques, such as bootstrapping or testing (for instance Mann-Whitney test or Fisher's test).
In particular embodiments, the score classifies the recipients into at least four distinct classes of risk of allograft loss, notably very high score, high risk, intermediate risk and low risk group.
In some embodiments, the algorithm is implemented on a computer using well-known computer processors, memory units, storage devices, computer software, and other components. Typically, the computer contains a processor, which controls the overall operation of the computer by executing computer program instructions which define such operation. The computer program instructions may be stored in a storage device (e.g., magnetic disk) and loaded into memory when execution of the computer program instructions is desired. The computer also includes other input/output devices that enable user interaction with the computer (e.g., display, keyboard, mouse, speakers, buttons, etc.). One skilled in the art will recognize that an implementation of an actual computer could contain other components as well.
In some embodiments, the algorithm is implemented using computers operating in a client-server relationship. Typically, in such a system, the client computers are located remotely from the server computer and interact via a network. The client-server relationship may be defined and controlled by computer programs running on the respective client and server computers. In some embodiments, the results may be displayed on the system for display, such as with LEDs or an LCD. Accordingly, in some embodiments, the algorithm can be implemented in a computing system that includes a back-end component, e.g., as a data server, or that includes a middleware component, e.g., an application server, or that includes a front-end component, e.g., a client computer having a graphical user interface or a Web browser through which a user can interact with an implementation, or any combination of one or more such back-end, middleware, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication, e.g., a communication network. Examples of communication networks include a local area network (“LAN”) and a wide area network (“WAN”), e.g., the Internet. The computing system can include clients and servers. A client and server are generally remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other.
In some embodiments, the algorithm is implemented within a network-based cloud computing system. In such a network-based cloud computing system, a server or another processor that is connected to a network communicates with one or more client computers via a network. A client computer (e.g. a mobile device, such as a phone, tablet, or laptop computer) may communicate with the server via a network browser application residing and operating on the client computer, for example. A client computer may store data on the server and access the data via the network. A client computer may transmit requests for data, or requests for online services, to the server via the network. The server may perform requested services and provide data to the client computer(s). The server may also transmit data adapted to cause a client computer to perform a specified function, e.g., to perform a calculation, to display specified data on a screen, etc. For instance, the physician may register the parameters (i.e. input data) on, which then transmits the data over a long-range communications link, such as a wide area network (WAN) through the Internet to a server with a data analysis module that will implement the algorithm and finally return the output (e.g. score) to the mobile device.
In some embodiments, the output results can be incorporated in a Clinical Decision Support (CDS) system. These output results can be integrated into an Electronic Medical Record (EMR) system.
In other words, the interaction between a computer program product and the system enables to carry out the method of the invention. The method of the invention is thus a computer-implemented method.
This means that the method is, at least partly computer-implemented.
In particular, each step can be computer-implemented provided some steps are achieved by receiving data.
The system is a desktop computer. In variant, the system is a rack-mounted computer, a laptop computer, a tablet computer, a Personal Digital Assistant (PDA) or a smartphone.
In specific embodiments, the computer is adapted to operate in real-time and/or is an embedded system, notably in a vehicle such as a plane.
In the present case, the system comprises a calculator, a user interface and a communication device.
The calculator is electronic circuitry adapted to manipulate and/or transform data represented by electronic or physical quantities in registers of the system X and/or memories in other similar data corresponding to physical data in the memories of the registers or other kinds of displaying devices, transmitting devices or memory devices.
As specific examples, the calculator comprises a monocore or multicore processor (such as a central processing unit (CPU), a graphics processing unit (GPU), a microcontroller and a Digital Signal Processor (DSP)), a programmable logic circuitry (such as an Application Specific Integrated Circuit (ASIC), a Field Programmable Gate Array (FPGA), a Programmable Logic Device (PLD) and programmable logic arrays (PLA)), a state machine, gated logic and discrete hardware components.
The calculator comprises a data-processing unit which is adapted to process data, notably by carrying out calculations, memories adapted to store data and a reader adapted to read a computer readable medium.
The user interface comprises an input device and an output device.
The input device is a device enabling the user of the system to input information or command to the system.
In the present case, the input device is a keyboard. Alternatively, the input device is a pointing device (such as a mouse, a touch pad and a digitizing tablet), a voice-recognition device, an eye tracker or a haptic device (motion gestures analysis).
The output device is a graphical user interface, that is a display unit adapted to provide information to the user of the system.
In the present case, the output device is a display screen for visual presentation of output. In other embodiments, the output device is a printer, an augmented and/or virtual display unit, a speaker or another sound generating device for audible presentation of output, a unit producing vibrations and/or odors or a unit adapted to produce electrical signal.
In a specific embodiment, the input device and the output device are the same component forming man-machine interfaces, such as an interactive screen.
The communication device enables unidirectional or bidirectional communication between the components of the system. For instance, the communication device is a bus communication system or a input/output interfaces.
The presence of the communication device enables that, in some embodiments, the components of the calculator be remote one from another.
The computer program product comprises a computer readable medium.
The computer readable medium is a tangible device that can be read by the reader of the calculator.
Notably, the computer readable medium is not a transitory signal per se, such as radio waves or other freely propagating electromagnetic waves, such as light pulses or electronic signals.
Such computer readable storage medium is, for instance, an electronic storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device or any combination thereof.
As a non-exhaustive list of more specific examples, the computer readable storage medium is a mechanically encoded device such a punchcards or raised structures in a groove, a diskette, a hard disk, a read-only memory (ROM), a random access memory (RAM), an erasable programmable read-only memory (EROM), electrically erasable and programmable read only memory (EEPROM), a magnetic-optical disk, a static random access memory (SRAM), a compact disc read-only memory (CD-ROM), a digital versatile disk (DVD), a memory stick, a floppy disk, a flash memory, a solid state drive disk (SSD) or a PC card such as a Personal Computer Memory Card International Association (PCMCIA).
A computer program is stored in the computer readable storage medium. The computer program comprises one or more stored sequence of program instructions.
Such program instructions when run by the data-processing unit, cause the execution of steps of the method of the invention.
For instance, the form of the program instructions is a source code form, a computer executable form or any intermediate forms between a source code and a computer executable form, such as the form resulting from the conversion of the source code via an interpreter, an assembler, a compiler, a linker or a locator. In variant, program instructions are a microcode, firmware instructions, state-setting data, configuration data for integrated circuitry (for instance VHDL) or an object code
Program instructions are written in any combination of one or more languages, such as an object oriented programming language (FORTRAN, C++, JAVA, HTML), procedural programming language (language C for instance).
Alternatively, the program instructions are downloaded from an external source through a network as it is notably the case for applications. In such case, the computer program product comprises a computer-readable data carrier having stored thereon the program instructions or a data carrier signal having encoded thereon the program instructions.
In each case, the computer program product comprises instructions which are loadable into the data-processing unit and adapted to cause execution of the method of the invention when run by the data-processing unit. According to the embodiments, the execution is entirely or partially achieved either on the system, that is a single computer, or in a distributed system among several computers (notably via cloud computing).
According to embodiment, the above-described method is implemented in many ways, notably using hardware, software or a combination thereof. In particular, each step is implemented by a module adapted to achieve the step or computer instructions adapted to cause the execution of the step by interaction with the system or a specific apparatus comprising the system.
It should also be noted that two steps in succession may, in fact, be executed substantially concurrently or in a reverse order depending on the considered embodiments.
The method as disclosed herein is useful for identifying patients with a high risk of allograft loss. In particular, the method of the present invention is particularly suitable for selecting a therapeutic regimen or determining if a certain therapeutic regimen is more appropriate for a patient identified as having a high risk of allograft loss. Typically, this regimen treatment consists of triple therapy regimen comprising a corticosteroid plus a calcineurin inhibitor (e.g. Ciclosporin, Tacrolimus) and an anti-proliferative agent (e.g. Azathioprine, Mycophenolic acid) may be used. mTOR inhibitors (e.g. Sirolimus, Everolimus) also may be used. Anti-CD25 antibodies may be used such as basiliximab. Reducing the level and the production of the DSA and/or protecting the allograft may be achieved using any suitable medical means known to those skilled in the art. In some embodiments, such reduction and protection comprise a therapeutic intervention with the subject such as administration of antithymomcy globulin (ATG), administration of B cell depleting antibodies, administration of proteasome inhibitor (bortezomib), intravenous administration of immunoglobulins, plasmapheresis, administration of anti-C5 antibodies (e.g. eculizumab) and splenectomy. Typical B cell depleting antibodies include but are not limited to anti-CD20 monoclonal antibodies [e.g. Rituximab (Roche), Ibritumomab tiuxetan (Bayer Schering), Tositumomab (GlaxoSmithKline), AME-133v (Applied Molecular Evolution), Ocrelizumab (Roche), Ofatumumab (HuMax-CD20, Gemnab), TRU-015 (Trubion) and IMMU-106 (Immunomedics)1, an anti-CD22 antibody [e.g. Epratuzumab, Leonard et al., Clinical Cancer Research (Z004) 10: 53Z7-5334], anti-CD79a antibodies, anti-CD27 antibodies, or anti-CD19 antibodies (e.g. U.S. Pat. No. 7,109,304), anti-BAFF-R antibodies (e.g. Belimumab, GlaxoSmithKline), anti-APRIL antibodies (e.g. anti-human APRIL antibody, ProSci inc.), and anti-IL-6 antibodies [e.g. previously described by De Benedetti et al., J Immunol (2001) 166: 4334-4340 and by Suzuki et al., Europ J of Immunol (1992) 22 (8) 1989-1993, fully incorporated herein by reference]. AMR can also require blood exchanges (Mg, plasmatic exchanges) to remove antibodies present in the recipient circulating compartment and targeting the graft. Reciprocally, where the recipient is predicted a low risk of allograft loss, the immunosuppressive therapy can be reduced in order to diminish the potential for drug toxicity.
In some embodiments, the method of the present invention may be used to identify patients in need of frequent follow-up by a physician or clinician to monitor the therapeutic regimen. In some embodiments, a patient can be monitored using the method as disclosed herein, and if on a first (i.e. initial) testing the patient is identified as having a high risk of allograft rejection, the patient can be administered an appropriate therapeutic regiment, and on a second testing (i.e. follow-up testing), the patient is identified as having low risk of allograft loss, the patient can be administered with a therapeutic regiment at a maintenance dose.
Thus, the method of the present invention is particularly suitable for discriminating responder from non-responder. As used herein the term “responder” in the context of the present disclosure refers to a subject that will achieve a response, i.e. the risk of allograft loss does show a reduction. A non-responder subject includes subjects for whom the risk of allograft loss does not show any reduction or improvement after the treatment.
Accordingly, the present invention further concerns a method for discriminating a responder recipient from a non-responder recipient to a given treatment regimen, said method comprising the steps of:
(i) implementing the prediction method of the invention on a recipient treated with said given treatment regimen,
(ii) if the recipient is predicted as having a high-risk of allograft loss, identifying the recipient as a non-responder recipient to said treatment regimen, or if the recipient is predicted as having a low risk of allograft loss, identifying the recipient as a responder recipient to said treatment regimen.
In some embodiments, screening patients for identifying patients having a high risk of allograft loss using the prediction method as disclosed herein is also useful to identify patients most suitable or amenable to be enrolled in clinical trial for assessing a therapy for management of allograft, which will permit more effective subgroup analyses and follow-up studies. Furthermore, the prediction method as disclosed herein can be suitable for monitoring patients enrolled in a clinical trial to provide a quantitative measure for the therapeutic efficacy of the therapy which is subject to the clinical trial.
The present invention thus also concerns a method of monitoring recipients enrolled in a clinical trial concerning a given therapy, said method comprising the step of implementing the prediction method of the invention, thereby providing a quantitative measure for the therapeutic efficacy of the therapy which is subject to the clinical trial.
Accordingly, the output of the algorithm (e.g. the score) can represent a suitable surrogate marker for use in a clinical trial for assessing the efficiency of a particular therapy.
Therefore, in a particular embodiment of the prediction method of the invention, the output of the algorithm (e.g. the score) constitutes a surrogate marker for use in a clinical trial for assessing the efficiency of a particular therapy.
The invention will be further illustrated by the following figures and examples. However, these examples and figures should not be interpreted in any way as limiting the scope of the present invention.
Study Design and Participants
Derivation cohort. The derivation cohort consisted of 4,000 consecutive patients over 18 years of age who were prospectively enrolled at the time of kidney transplantation from a living or deceased donor at Necker Hospital (n=1,473), Saint-Louis Hospital (n=928), Foch Hospital (n=714), and Toulouse Hospital (n=885) between Jan. 1, 2005, and Jan. 1, 2014, in France. The clinical data were collected from each centre and entered into the Paris Transplant Group database (French data protection authority (CNIL) registration number: 363505). All data were anonymised and prospectively entered at the time of transplantation, at the time of posttransplant allograft biopsies and at each transplant anniversary using a standardised protocol to ensure harmonisation across study centres. Data from the derivation cohort were submitted for an annual audit to ensure data quality (See EXAMPLE 2 for detailed data collection procedures). Data were retrieved from the database on March 2018. The institutional review boards of the Paris Transplant Group participating centres approved the study. All patients provided written informed consent at the time of transplantation.
Validation cohorts. External validation was conducted on 3,557 kidney transplant recipients from a living or a deceased donor over 18 years of age and representing all eligible patients for posttransplant risk evaluation (i.e., undergoing allograft biopsy as part of the standard of care of each centre with adequate biopsy according to the Banff criteria) from six centres: 2,129 recipients recruited in Europe and 1,428 recipients recruited in North America between 2002 and 2014. The European centres included Hopital Hotel Dieu, Nantes, France (n=632), Hospices Civils, Lyon, France (n=608), and the University Hospitals, Leuven, Belgium (n=889). The US centres included the Johns Hopkins Medical Institute, Baltimore, Md. (n=580), the Mayo Clinic, Rochester, Minn. (n=556), and the Virginia Commonwealth University School of Medicine, Richmond, Va. (n=292). Data sets from the validation centres were prospectively collected as part of routine clinical practice and entered in the centres' databases in compliance with local and national regulatory requirements and sent anonymised to the Paris Transplant Group.
In France, the transplantation allocation system followed the rules of the French National Agency for Organ Procurement (Agence de la Biomédecine). Centres outside of France followed the rules of the Eurotransplant allocation system (Leuven),(18) whereas US centres (Johns Hopkins Hospital, Mayo Clinic and Virginia) followed the rules of the US Organ Procurement and Transplantation System.(19)
Additional external validation cohort. Additional external validation was conducted in kidney transplant recipients previously recruited in three registered and published phase II and III clinical trials: a randomised, open-label, multicentre trial that compared a cyclosporine-based immunosuppressive regimen to an everolimus-based regimen in kidney recipients (Certitem, NCT01079143); a randomised, multicentre, double-blind, placebo-controlled trial that investigated the efficacy of rituximab in kidney recipients with acute antibody-mediated rejection (Rituxerah, EudraCT 2007-003213-13); and a randomised, double-blind placebo-controlled single-centre trial that investigated the efficacy of bortezomib in kidney recipients with late antibody-mediated rejection (Borteject, NCT01873157).(20-22) The details of the clinical trials depicting the population characteristics, study design, inclusion criteria and interventions are provided in Table 4.
Candidate Predictors
Posttransplant risk evaluation times. Risk evaluation after transplantation was conducted at the time of allograft biopsy performed for clinical indication or as per protocol, which was performed after transplantation according to the centres' practices. In patients with multiple biopsies, risk evaluation was performed using the date of the first biopsy. The distribution of posttransplant risk evaluation times is provided in
Patient risk evaluation after transplant comprised demographic characteristics (including recipient comorbidities, age, gender and transplant characteristics), biological parameters (including kidney allograft function, proteinuria, and circulating anti-HLA antibody specificities and levels), and allograft pathology data (including elementary lesion scores and diagnoses), All these factors are commonly and routinely collected in kidney transplant centres worldwide.
See EXAMPLE 2 for the list of all prognostic determinants assessed from the derivation cohort.
Measurements performed at the time of risk evaluation. Kidney allograft function was assessed by the glomerular filtration rate estimated by the Modification of Diet in Renal Disease Study equation (eGFR) and proteinuria level using the protein/creatinine ratio in the derivation and validation cohorts. Circulating donor-specific antibodies against HLA-A, HLA-B, HLA-Cw, HLA-DR, HLA-DQ and HLA-DP were assessed using single-antigen flow bead assays in the derivation cohort (see EXAMPLE 2) and according to local centre practice in the validation cohorts. Kidney allograft pathology data, including elementary lesion scores and diagnoses, were recorded according to the Banff classification in the derivation and validation cohorts (see EXAMPLE 2). All the measurements (eGFR, proteinuria, histopathology and circulating anti-HLA DSA) were performed on the day of risk evaluation.
Outcome
The outcome of interest was allograft loss defined as a patient's definitive return to dialysis or preemptive kidney retransplantation. This outcome was prospectively assessed in the derivation and validation cohorts at each transplant anniversary up to Mar. 31, 2018.
Patient death was considered as a competing event (see EXAMPLE 2).
Missing Data
A total of 59 patients (0.01%) were excluded from the final model due to at least one missing data point.
Statistical Analysis
We followed the TRIPOD (Transparent Reporting of a Multivariable Prediction Model for Individual Prognosis or Diagnosis) statement for reporting multivariable prediction model development and validation.(23)
Continuous variables are described using means and standard deviations (SDs) or median and the interquartile range. We compared means and proportions between groups using Student's t-test, analysis of variance (ANOVA) (Mann-Whitney test for MFI) or the chi-square test (or Fisher's exact test if appropriate). Graft survival was estimated using the Kaplan-Meier method. The duration of follow-up started with the patient risk evaluation (starting point) up to the date of kidney allograft loss, or at the end of the follow-up (Mar. 31, 2018). For patients who died with a functioning allograft, allograft survival was censored at the time of death as a surviving or functional allograft.(24) A competing risk approach was applied to consider the potential competition of patient death with kidney allograft failure (see EXAMPLE 2).(25)
In the derivation cohort, the associations between allograft failure and clinical, histologic, functional, and immunologic factors measured at the patient risk evaluation (see above) were assessed using univariable Cox regression analyses. Hazard proportional assumptions were tested using the log graphic method. The factors identified in these analyses were thereafter included in a final multivariable model.
The internal validity of the final model was confirmed using a bootstrap procedure, which involved generating 1,000 datasets derived from resampling the original dataset and permitting the estimation of the biased corrected 95% CI and the accelerated bootstrap (BCA) HR.(26)
The centre effect was tested in stratified analyses. Potential nonlinear relationships between continuous predictors and graft loss were first investigated using restricted cubic splines and fractional polynomial methods (see EXAMPLE 2).
The accuracy of the prediction model was assessed based on its discrimination ability and calibration performance. The discrimination ability (i.e., the ability to separate patients with different prognoses) of the final model was evaluated using Harrell's concordance index (C-index) (see EXAMPLE 2).(27) One thousand random samples of the population were used to derive the 95% confidence intervals (CIs) for the C-index. Calibration and goodness of fit (the ability to provide unbiased survival predictions in groups of similar patients) were assessed based on a visual examination of the calibration plots and tested with an extension of the Hosmer-Lemeshow test for survival data. Net reclassification improvement for censored survival data was computed using the SurvIDINRI package in R. (28, 29) The external validity of the final model was thereafter evaluated in the external validation cohorts, including discrimination tests and model calibration as mentioned above.
A risk prediction score (“integrative box risk prediction score, iBox”) was calculated for each patient according to the (3-regression coefficients estimated from the final multivariable Cox model and normalised to a range between 0 and 5 (see EXAMPLE 2). To obtain a reasonable spread of risk, we chose to work on five prognostic risk groups. Cox's method was applied to determine optimal nonarbitrary cut-off points to define five risk groups. (30)
All analyses were performed using R (version 3.2.1, R Foundation for Statistical Computing). Values of p<0.05 were considered significant, and all tests were 2-tailed. Details regarding the interpretation of important statistical concepts are given in EXAMPLE 2.
Characteristics of the Derivation and Validation Cohorts
The derivation cohort (n=4,000) and the two validation cohorts (n=3,557) comprised a total of 7,557 participants. The characteristics of the derivation and validation cohorts (overall, European and North American validation cohorts) as well as the transplant procedures, policies and allocation systems are detailed in Table 1 and Tables 5, 6, and 7. The distribution of the time of posttransplant risk evaluation is provided in
Prediction of Kidney Allograft Failure in the Derivation Cohort
We first investigated the prognostic factors measured at the time of posttransplant risk evaluation that were associated with long-term kidney allograft failure in a univariable analysis. These factors included recipient demographics, transplant characteristics, allograft functional parameters, immunological parameters, and allograft histopathology (Table 2A). In the multivariable analysis, the following independent predictors of long-term allograft failure were identified: i) time of posttransplant risk evaluation (p=0.005); ii) allograft functional parameters, including estimated glomerular filtration rate (eGFR; p<0.001) and proteinuria (logarithmic transformation, p<0.001); iii) allograft histological parameters, including interstitial fibrosis and tubular atrophy (p=0.031), microcirculation inflammation defined by glomerulitis and peritubular capillaritis (p=0.001), interstitial inflammation and tubulitis (p=0.014) and transplant glomerulopathy (p=0.004); and iv) recipient immunological profile as defined by the presence and level of the immunodominant circulating anti-HLA donor-specific antibodies (p<0.001; Table 2B). To test the centre effect, we stratified the final multivariable model by transplant centres and confirmed that the eight prognostic parameters identified in the primary analysis remained independently associated with allograft survival (Table 8). Using competing risk regression models, we confirmed that the allograft survival analyses performed in the final model were not affected by competition with patient death (see EXAMPLE 2, and
The prognostic score, named iBox, was calculated for each patient according to the (3-regression coefficients estimated from the final multivariable Cox model and normalised to a range between 0 and 5 (see EXAMPLE 2). The population was divided into five risk groups with an increasingly higher risk of graft loss corresponding to the following cut-off points: iBox risk strata 1 (n=1,104): <1.805; iBox risk strata 2 (n=1,149): 1.805-2.265; iBox risk strata 3 (n=896): 2.265-2.705; iBox risk strata 4 (n=551): 2.705-3.275; and iBox risk strata 5 (n=241): >3.275. This stratification achieved a clear separation of the Kaplan-Meier curves, defining five subgroups of patients with distinct long-term allograft prognoses, with 7-year post-risk evaluation allograft survival rates of 96% (95% CI: 94 to 97), 91% (95% CI: 89 to 93), 82% (95% CI: 79 to 85), 59% (95% CI: 54 to 65), and 33% (95% CI: 26 to 41) in strata 1, 2, 3, 4 and 5, respectively (
Prediction Model Performance in the Internal and External Validation Cohorts
We first internally validated the final multivariable model via a bootstrapping procedure with 1,000 samples from the original dataset of the derivation cohort (EXAMPLE 2). Using this approach, we confirmed 1) the robustness of the final multivariable model (bias-corrected HRs and 95% CIs, Table 2B); 2) the successful discrimination ability at 3, 5 and 7 years (C-index: 0.83, 95% bootstrap percentile CI=0.81 to 0.86; 0.82, 95% bootstrap percentile CI=0.80 to 0.84; 0.81; 95% bootstrap percentile CI=0.79 to 0.83, respectively) of the model; and 3) the accurate calibration at 3, 5 and 7 years (p=0.85, p=0.65 and p=0.36, respectively) (
We then used several independent validation cohorts and confirmed the transportability of the iBox risk score in these geographically distinct cohorts. Overall, we demonstrated good discrimination performance in the external validation cohorts with a C statistic of 0.81 in Europe (95% bootstrap percentile CI=0.78 to 0.84) and 0.80 in the US (95% bootstrap percentile CI=0.76 to 0.84). The calibration plots showed optimal agreement between the iBox risk score-predicted probabilities of allograft survival at 3, 5 and 7 years after risk evaluation and actual kidney allograft survival (
Performance of the iBox Risk Prediction Score in Therapeutic Randomised Controlled Clinical Trials
We tested the performance of the iBox risk prediction score in 3 registered and published phase II and III clinical trials.(20-22). The details of the clinical trials depicting the population, intervention, clinical scenario and follow-up times are presented in Table 4. We calculated the iBox risk prediction scores of all patients included in the trials and compared those with the actual allograft failures. The iBox risk prediction score applied in the three trials revealed accurate discrimination overall (C-index 0.87; 95% bootstrap percentile CIs=0.82 to 0.92). The calibration plot showed an optimal agreement between the risk prediction score based on predicted allograft loss and the actual observations of kidney allograft loss (
Sensitivity Analyses
Various sensitivity analyses were performed to test the robustness and generalisability of the iBox risk score in different clinical scenarios and subpopulations.
Added Value of the iBox Integrative Risk Prediction Score Compared to Conventional Allograft Function Monitoring (eGFR/Proteinuria) and Generation of an Abbreviated Functional iBox Score.
We tested the added value of the iBox risk prediction score over the conventional allograft monitoring model based on eGFR and proteinuria assessments. We demonstrated that the iBox risk score was superior in terms of prediction capability than a restricted model including eGFR and proteinuria (C-index=0.73; 95% bootstrap percentile CI=0.71 to 0.75, p-value <0.0001 as compared with the full iBox model). This was further demonstrated by a continuous net reclassification improvement (cNRI) of 0.228 for the iBox model compared to that of the functional model (95% CI, 0.174 to 0.290, p<0.0001). To account for potentially different medico-economic contexts limiting the availability of allograft biopsies, we are providing in this study an abbreviated iBox score based on clinical-functional parameters (
Added Value of the iBox Risk Prediction Score Compared to Risk Scores Previously Reported in the Literature
We performed a systematic review (see EXAMPLE 3) and compared the iBox risk prediction score to previously published risk scores assessing long-term allograft outcomes and demonstrated that the iBox prediction score outperformed other risk scores (see EXAMPLE 3).
Prediction Model Performance using Histological Diagnoses instead of the Banff International Classification Histological Lesion Grading
When histological diagnoses were included in the multivariable model instead of histological lesions graded according to the international Banff classification, antibody-mediated rejection (AMR) (p<0.001), T-cell mediated rejection (TCMR, p=0.045), primary nephropathy recurrence (p=0.003) and BK virus nephropathy (BKVAN, p=0.050) showed significant and independent associations with allograft failure. In this model, the set of non-histological predictors of allograft failure identified in the primary analyses remained unchanged (hazard ratios are shown for each parameter in Table 10). The discrimination ability of the histological diagnosis-based model revealed a C-index of 0.76 (95% bootstrap percentile CI=0.74 to 0.81).
iBox Performance when Applied at the Time of Clinically Indicated Biopsies vs Protocol Biopsies
We tested and confirmed the performance of the iBox risk prediction score when risk evaluation started at the time of clinically indicated allograft biopsies performed at any time after transplantation (n=1,598, 40%), as well as at the time of 1-year protocol biopsies (n=2,402, 60%; Table 3).
Similarly, the iBox risk score demonstrated accurate discrimination ability for long term allograft loss when risk evaluation started before 1-year post transplant or after 1-year post transplant (average post-transplantation time of 0.89±0.23 years and 2.31±1.66 years respectively; Table 3).
IBox Assessed in other Clinical Scenarios and Subpopulations
Finally, we confirmed the performance of the iBox risk prediction score when applied in different subpopulations and clinical scenarios including i) living and deceased donors, ii) according to recipient's ethnicity, iii) in highly sensitised (high immunological risk) and non-highly sensitised (low immunological risk) recipients, and iv) patient receiving an induction by anti-iL2 receptor or anti-thymocyte globulin (Table 3). When parameters assessed at the time of transplant (such as HLA mismatches), recipient blood pressure at the time of risk assessment (log scale), and calcineurin inhibitor through blood level at the time of risk assessment were forced in the risk prediction score, there was no significant improvement in its prognostic performance (Table 3).
The iBox, a risk prediction score combining allograft functional, histological, and immunological parameters together with HLA antibody profiling, showed good performance in predicting the risk of long-term kidney allograft failure. We demonstrated the generalisability of the iBox risk prediction score by showing its external validity in six geographically distinct cohorts recruited in Europe and in the US with distinct allocation systems, patient characteristics and management practices. The iBox risk prediction score also demonstrated its accuracy when measured at different times post-transplantation, which permits to update the score based on new events that patient might encounter in their long-term course. We also demonstrated the added value of the iBox risk prediction score over a conventional allograft monitoring model that includes eGFR and proteinuria assessment and showed that the iBox risk prediction score outperformed other available risk scores applied in kidney transplant patients. Last, we confirmed the predictive accuracy of the risk score in the data issued from three published randomised therapeutic trials covering different clinical scenarios encountered after transplantation, further enhancing its value as a potential surrogate endpoint in transplantation.(20-22) Overall, the predictor variables used in the iBox risk prediction score are easily available after transplantation in most centres worldwide, making it feasible for implementation in routine clinical practice. To account for potential different medico-economic contexts limiting the availability of allograft biopsies, we also provide in this study an abbreviated score based on clinical-functional parameters.
Current prognostic scores implemented in clinical practice in transplant medicine mostly address the prediction of allograft survival at the time of transplantation; thus, their use is limited to allograft allocation because they do not inform posttransplant clinical decision making and patient monitoring.(31) The few attempts of developing posttransplant prognostic scores have failed to provide useful tools for transplant clinicians. According to a systematic review without date restrictions for publications up to Jul. 25, 2018, for allograft survival scoring systems among kidney transplant recipients (see EXAMPLE 3), no study has developed and externally validated a posttransplant prognostic score usable at any time after transplantation. The main limitations to achieve a robust and validated scoring system rely on multiple factors including the insufficient data quality of the previously studied cohorts and the fact that no registry or database system has been primarily designed to address the specific aspect of prognostication. An even more important aspect is external validation in different populations, which prompted us to conduct a large external validation from multiple centres worldwide. Despite some expected loss of discriminative performance, models are typically considered useful for clinical decision making when the C-statistic is greater than 0.70 and strong when the C-statistic exceeds 0.80, suggesting that the iBox risk prediction score could support decision making.(32) Compared to prognostication systems in other fields such as oncology (e.g., locally advanced pancreatic cancer and metastatic colonic cancer), the C-index is typically closer to 0.60 or 0.70.(33) Taken together, these results confirm not only the robustness and validity of the iBox risk prediction score but also its generalisability to other transplant cohorts with different kidney allocation systems, donor and recipient profiles, and distinct patient management and healthcare environments.
In this study, we demonstrated that the iBox risk prediction score outperformed the current gold standard (eGFR and proteinuria) for the monitoring of kidney recipients. In particular, compared to prior attempts at developing a prognostication system, we found that allograft histological lesions such as microcirculation inflammation, interstitial inflammation-tubulitis (reflecting active rejection process) and atrophy-fibrosis, and transplant glomerulopathy (reflecting chronic allograft damage), in addition to measuring allograft functional parameters and recipient antibody profiles, improved the overall discrimination capacity of the model and that a multidimensional risk prediction score performs better than its individual components. This risk prediction score reflects the main patterns of allograft deterioration leading to failure, represented by alloimmune processes and allograft scarring.(34) Two other prognostic scores have attempted to combine several transplant diagnostic dimensions, including allograft function and pathology and alloantibodies; however, these scores were outperformed by the iBox risk prediction score.(16, 35)
Importantly, our results and the parameters included in the final model reinforce the potential of the iBox to be implemented into contemporary clinical practice by using automated approaches within electronic medical record systems (an online available electronic risk calculator is provided at http://www.paristransplantgroup.org and examples are provided in EXAMPLE 4).
In addition, the combination of major drivers of allograft failure in the iBox risk prediction score were allowed to evaluate early the effect of clinical interventions on long-term allograft outcomes. In this study, we tested and validated the iBox risk prediction score in the setting of therapeutic clinical trials covering different clinical scenarios and demonstrated accurate performance overall. We found that the prediction of allograft failure assessed by the iBox score accurately fits with the actual graft failures observed in these trials at 5 years after risk evaluation. Importantly, the accuracy of the iBox risk prediction score was conserved regardless of the therapeutic intervention and population from those trials, with accurate performance in the Certitem (NCT01079143) calcineurin inhibitor minimisation trial (22) and rejection treatment trials (EudraCT 2007-003213-13; NCT01873157).(20, 21) This finding reinforced the potential of the iBox risk prediction score for defining a valid surrogate endpoint. Indeed, in the present study, a well-validated, strong and robust association existed between the surrogate and the true endpoint, and this association was consistent across different treatment settings.
Regarding the limitations of this study, we acknowledge that emerging predictors posttransplant might be missing in our model. Despite the already high performance achieved by the iBox risk prediction score, future studies should evaluate the added value of new non-invasive biomarkers or genetic factors in addition to those presently reported regarding discriminative capability, generalisability and overcoming the need for an invasive procedure (kidney allograft biopsy). Another limitation is that information regarding the drug adherence of single patients was lacking in our dataset. Although nonadherence is a major risk factor for graft failure, it is inherently difficult to capture, especially at a population level.(34) Notwithstanding that the iBox risk prediction score was primarily generated using a large, prospective, unselected cohort, a prospective validation of the iBox in daily clinical practice remains desirable. Finally, despite the validation of the iBox risk prediction score in an interventional setting, future trials are needed to compare whether a strategy based on a systematic risk evaluation vs. an empirical approach might improve clinical management.
We developed and validated the iBox risk prediction score, which accurately predicts allograft failure after kidney transplantation. We demonstrated its generalisability and transportability across centres worldwide and its performance in therapeutic clinical trials. The iBox risk prediction score provides an accurate but simple strategy that can be easily implemented to stratify patients into clinically meaningful risk groups and that can be time-updated after transplant which may help guide patient monitoring in everyday practice and stratify patients in future clinical trials.
Data Collection Procedures
All data from Paris-Necker, Paris-Saint Louis, Foch and Toulouse hospitals were extracted from the prospective Paris Transplant Group Cohort data cohort (CNIL, Registration number: 363505, validated on the 8 of Jun. 2004). The database networks have been approved by the National French Commission for bioinformatics data and patient liberty and codes were used to ensure strict donor and recipient anonymity and blind access. Informed consent was obtained from the participants at the time of transplantation. The data are computerised in real time and at the time of transplantation, at the time of post-transplant allograft biopsies and at each transplant anniversary and are submitted for an annual audit.
Independent Validation Cohorts
In the European validation cohort, the French data from the Lyon, and Nantes Hospitals for donors and recipients were extracted from the DIVAT clinical prospective cohort (official website: www.divat.fr) and from the French national agency database CRISTAL (official website: https://www.sipg.sante.fr/portail/). The Belgian data and data from the North-American validation cohort were collected as part of routine clinical practice and entered in centres' databases in compliance with local and national regulatory requirements. They were sent anonymised to the Paris Transplant Group.
Prognostic Parameters Prospectively Collected and Assessed in the Derivation Cohort
Baseline Recipient's Characteristics:
2. Recipient's gender
3. Recipient's height
4. Recipient's weight
5. Previous transplantation
6. Delay between dialysis and transplantation
7. Cause of end stage renal disease
8. ABO blood group
9. HLA genotype
10. CMV serology
11. HCV serology
12. HBV serology
13. HIV serology
Baseline Donor's Characteristics:
15. Donor's gender
16. Donor's height
17. Donor's weight
18. Type of donor: deceased vs living
19. Cause of donor's death
20. Double transplantation
21. History of hypertension
22. History of diabetes
23. ECD status
24. Serum creatinine
25. ABO blood group
26. HLA genotype
27. CMV serology
28. HCV serology
29. HBV serology
30. HIV serology
Immunological Characteristics at the Time of Transplantation:
31. HLA mismatches A
32. HLA mismatches B
33. HLA mismatches Cw
34. HLA mismatches DQ
35. HLA mismatches DR
36. HLA mismatches DP
37. Anti-HLA DSA at the time of transplantation
38. MFI of the anti-HLA DSA at the time of transplantation
39. cPRA
Transplant Characteristics:
40. Cold ischemia time
41. Delayed graft function
42. Induction treatment with anti-thymocyte globulin
43. Induction treatment with basiliximab
44. Steroid dose
Immunological Data at the Time of Risk Assessment (Luminex SA Assessment A, B, C, DP, DQ, DR)
46. MFI of immunodominant anti-HLA DSA
Histological Data According to the Banff Classification:
47. g Banff score
48. ptc Banff score
49. t Banff score
50. i Banff score
51. cg Banff score
52. v Banff score
53. mm Banff score
54. ci Banff score
55. ct Banff score
56. IFTA Banff score
57. cv Banff score
58. ah Banff score
59. C4d ptc deposition
61. Polyomavirus-associated nephropathy
62. ABMR status
63. TCMR status
64. Borderline category
Follow-Up Variables:
65. Episodes of pyelonephritis
66. Immunosuppression treatment
67. Type of treatment: calcineurin inhibitors, mycophenolate mofetil, mTOR inhibitors or belatacept
68. CNI blood through level at M12 and every year
69. Steroid dose at M12 and every year
70. Rejection therapy (e.g., steroid, plasma exchange, intravenous immunoglobulin)
71. CMV prophylaxis
72. BK viral load at M12 and every year
73. CMV viral load at M12 and every year
74. Allograft function at M12 and every year
75. Proteinuria at M12 and every year
76. Patient date and cause of allograft loss
77. Patient date and cause of death
Detection and Characterisation of Donor-specific Anti-HLA Antibodies
All patients were tested for the presence of circulating anti-HLA donor-specific antibodies (DSAs) at the time of patient risk evaluation. The presence of circulating DSAs against HLA-A, HLA-B, HLA-Cw, HLA-DR, HLA-DQ and HLA-DP was retrospectively determined using single-antigen flow bead assays (One Lambda, Inc., Canoga Park, Calif., USA) on a Luminex platform. Beads with a normalised mean fluorescence intensity (MFI), a measure of donor-specific antibody strength, of greater than 500 units were judged as positive as previously described. HLA typing of the transplant recipients and donors was performed using an Innolipa HLA Typing Kit (Innogenetics, Ghent, Belgium). In the validation cohorts, HLA genotyping and HLA antibody profiling were performed according to local centre practice.
Kidney Allograft Phenotypes at Time of Risk Assessment
In the derivation cohort, allograft biopsies were scored and graded from 0 to 3 according to the updated Banff criteria for allograft pathology for the following histological factors: glomerular inflammation (glomerulitis), tubular inflammation (tubulitis), interstitial inflammation, endarteritis, peritubular capillary inflammation (capillaritis), transplant glomerulopathy, interstitial fibrosis, tubular atrophy, arteriolar hyalinosis and arteriosclerosis. Additional diagnoses provided by the biopsy (e.g., the diagnoses of primary disease recurrence, BK virus nephropathy) were recorded. The biopsy sections (4 μm) were stained with periodic acid-Schiff, Masson' s trichrome, and hematoxylin and eosin. C4d staining was performed via immunohistochemical analysis on paraffin sections using polyclonal human anti-C4d antibodies. Also, in the validation cohorts, the Banff criteria for the individual histological lesions were assessed in each biopsy included in the study.
Statistical Analysis Interpretation
Continuous Variables
When used as continuous variables in the Cox model, a potential non-linear relationship between predictors and allograft loss was first investigated using restricted cubic splines modelling. Secondly, a fractional polynomial method was applied to determine the best transformation for continuous variables. For donor age, recipient age, eGFR and HLA mismatches, a linear relationship with outcome was found to be a good approximation. A logarithmic transformation was necessary for proteinuria and time post-transplant.
Discrimination
The aim of discrimination is to distinguish between patients who experience an event and those who do not. The C-index estimates the proportion of all pairwise patient combinations from the sample data whose survival time can be ordered such that the patient with the highest predicted survival is the one who actually survived longer (discrimination). The C-index (0≤C≤1) is the probability of concordance between predicted and observed survival, with C-index=0.5 for random predictions and C-index=1 for a perfectly discriminating model.
Calibration
Calibration refers to the ability to provide unbiased survival predictions in groups of similar patients. It estimates how close the score-estimated risk is to the observed risk. A prediction model is considered “well calibrated” if the difference between predictions and observations in all groups of similar patients is close to 0 (perfect calibration). Any large deviation (p<0.1) indicates a lack of calibration.
Bootstrapping
Bootstrapping is the preferred simulation technique that was first described by Bradley Efron. The original dataset is a random sample of patients being representative of a general population. Bootstrapping means generating a large number of datasets, each of which with the same sample size as the original one, by resampling with replacement (i.e., a previously selected patient may be selected again).
Internal Validation
Internal validation is useful to obtain an honest estimate of the model performance for patients who are similar to those in the development sample and to indicate an upper limit to the expected performance in other settings. The bootstrap approach is the preferred technique to assess internal validity. The internal validity of the final model was confirmed using a bootstrap procedure, which involved generating 1,000 datasets derived from resampling the original dataset and permitted the estimation of the bias-corrected 95% CI and the accelerated bootstrap (BCA) HR.
External Validation
External validation may show different results from internal validation since many aspects may be different between settings, including selection of patients, definitions of variables, and diagnostic or therapeutic procedures. The strength of the evidence for the score validity is usually considered greater with a fully external validation (e.g., other investigators and centres).
Competing Risk by Death Analysis
We estimated cumulative incidence functions from competing risks data and compared the subdistribution for each cause across groups. We then assessed the effects of predictive factors (iBox risk strata) on the subdistribution of graft loss in a competing risks setting with death by fitting the proportional subdistribution hazard regression model described in the Fine and Gray method.
Construction of the Integrative Score Derived from the Final Multivariable Cox Model
α, β, χ, δ, ε, ζ, μ and φ: Cox-model beta coefficients for the corresponding parameters
A comprehensive search strategy was conducted through several databases (PubMed, Medline, Embase, Cochrane, and Scopus) without date restrictions for publications up to Jul. 25, 2018 for allograft survival scoring systems among kidney transplant recipients. We used the search terms “kidney transplantation”, “allograft survival” and “prognostic score”. Out of 460 articles identified, 11 were related to long-term allograft survival, 5 were externally validated and only 2 comprised immunological parameters. They are presented in Table 9 and compared with the iBox risk prediction score. The two studies identified: i) were not derived from patient cohorts with systematic monitoring and specific design towards risk stratification; ii) did not integrate a large spectrum of potential prognostic factors, iii) were not validated in multiple large cohorts worldwide with different transplant allocation systems and management practices, iv) were not validated in randomised controlled therapeutic clinical trials (RCTs).
Real-life patients for whom we used the iBox risk score to predict individual 3, 5 and 7-year-allograft survival. Patients #1 to #3 were from the iBox database reference set. Patient #4 was from the randomized controlled trial: RITUX ERAH Eudra CT 2007-003213-13.
Patient #1 Description
A 64-year-old male with membranoproliferative glomerulonephritis underwent a second preemptive kidney transplantation from an expanded-criteria deceased donor in 2013. The patient was sensitised (cPRA of 50%) without circulating anti-HLA DSA identified at the time of transplantation. Initial immunosuppressive regimen included anti-thymocyte globulin induction with corticosteroids, mycophenolate mofetil and tacrolimus.
Three months after transplantation, eGFR (MDRD) was 52 mL/min/1.73 m2 without proteinuria (0.05 g/g). The evaluation at one-year post-transplantation found an eGFR (MDRD) of 33 mL/min/1.73 m2. No circulating anti-HLA DSA nor proteinuria were detected. The biopsy revealed severe interstitial fibrosis and tubular atrophy was detected (IFTA Banff score=3) as well as a glomerulitis (g score=1). No other lesion was observed (ptc, c4d, cg, i, t scores=0).
iBoxPatient#1=βtime from transplant to risk evaluation*3+βeGFR*33+βProteinuria*log (0.05)+βDSA MFI*0+βg+ptc*1+βi+t*0+βcg*0+βIFTA*3
Patient #1 individual allograft survival probabilities at 3, 5 and 7-years are 94%, 91%, and 86% respectively (see
Patient #2 Description
A 39-year-old male patient with an obstructive uropathy underwent a first living-related donor kidney transplantation in 2012, with a cPRA at 0 at the time of transplantation. The immunosuppressive regimen consisted in basiliximab, corticosteroids, mycophenolate mofetil and tacrolimus.
At 15 months, the patient developed a de novo DSA (anti-DR4, MFI 8,244). At the time of dnDSA identification, the eGFR (MDRD) was 74 mL/min/1.73 m2 with a proteinuria of 1.51 g/g. A biopsy was performed with a g score=3, ptc score=2, C4d score=2, and transplant glomerulopathy score=1. No other lesion was observed (IFTA, i, t scores=0).
iBoxPatient#2=βtime from transplant to risk evaluation*15+βeGFR*74+βProteinuria*log(1.51)+βDSA MFI*3 (e.g. greater than 6,000 of MFI)+βg+ptc*5+βi+t*0+βcg*1+βIFTA*0
The iBox score projects the patient in the strata 3. The 3, 5 and 7-year probabilities of allograft survival are 86%, 78%, and 69% respectively (see
Patient #3 Description
A 45-year-old woman with an end-stage renal disease due to a type 1 diabetes underwent her first kidney transplantation with a standard criteria deceased donor in 2009. She was highly sensitised due to blood transfusions (cPRA=89%) but without detectable circulating anti-HLA DSA Immunosuppressive treatment included an induction therapy with anti-thymocyte globulin and a maintenance immunosuppressive regimen of corticosteroids, MMF and tacrolimus.
At 5 months post-transplant, eGFR (MDRD) was 62 mL/min/1.73 m2, a de novo DSA was detected (anti-DQ8, MFI 1,233) and a proteinuria was identified (0.19 g/g). An allograft biopsy revealed a mild interstitial fibrosis/tubular atrophy (IFTA score=1). The biopsy was free from other pathological lesion (g, ptc, c4d, cg, i, t Banff scores=0).
iBoxPatient#3, 1st score=βtime from transplant to risk evaluation*5+βeGFR*62+βProteinuria*log(0.19)+βDSA MFI*1 (e.g. between 500 and 3,000 of MFI)+βg+ptc*0+βi+t*0+βcg*0+βIFTA*1
The patient #3 individual allograft survival probabilities at 3, 5 and 7 years are 97%, 96%, and 94% respectively.
The same patient was then reevaluated 13 months post transplantation with a decreased eGFR at 43 ml/min/1.73 m2. The MFI of the anti-DQ8 dnDSA increased to 7′358. A new biopsy was performed showing a transplant glomerulopathy (cg score of 1). The other parameters were stable otherwise, when compared with the previous biopsy.
iBoxPatient#3, 2nd score=βtime from transplant to risk evaluation*13+βeGFR*43+βProteinuria*log(0.22)+βDSA MFI*3 (e.g. greater than 6,000)+βg+ptc*0+βi+t*0+βcg*1+βIFTA*1
The iBox prediction score for patient #2 is updated with 86%, 78%, and 68% individual allograft survival probabilities at 3, 5 and 7-years to respectively (see
Patient #4 Description (Rituxerah Trial Eudra CT 2007-003213-13)
A 56-year-old woman with tubulointerstitial nephropathy underwent a first kidney transplantation in 2011 (standard criteria deceased donor). At Day 0 no circulating anti-HLA DSA was detected and an induction with anti-thymocyte globulin was followed by an immunosuppression with corticosteroids, mycophenolate mofetil and calcineurin inhibitor. After 10 days, GFR was estimated at 48 mL/min/1.73 m2 without proteinuria.
At month 1 post-transplant, the patient presented with a decreased allograft function; eGFR of 25 mL/min/1.73 m2, a circulating de novo DSA (anti-B44, MFI 1,972), and a proteinuria of 2.07 g/g. A biopsy was performed and found an active ABMR (g2, ptc1, c4d3 according to Banff scoring system), with mild tubulitis (t score 1) and arteriolar hyalinosis (ah score 1). She was included in Rituxerah trial Eudra CT 2007-003213-13 in the placebo group (plasma exchange, intravenous immunoglobulin and steroid according to the protocol).
Below is the IBox evaluation at the time of patient inclusion:
IBoxPatient#4, 1st score=βtime from transplant to risk evaluation*1+βeGFR*25+βProteinuria*log(2.07)+βDSA MFI*1 (e.g. between 500 and 3000 of MFI)+βg+ptc*3+βi+t*1+βcg*0+βIFTA*0
The patient #4 individual allograft survival probabilities at the time of the therapeutic intervention were 59%, 43%, and 27% at 3, 5 and 7 years, respectively.
Six months after inclusion, the eGFR was of 37 mL/min/1.73 m2, proteinuria was 0.32 g/g of creatininuria and the previously identified anti-HLA DSA was undetectable. The biopsy found an acute borderline T-cell mediated rejection according to the Banff classification (i score 1 and t score 1), arteriosclerosis (cv score 1), mild arteriolar hyalinosis (ah score 1), glomerulitis score of 2 and interstitial fibrosis and tubular atrophy (IFTA score 3).
IBOXPatient#4, 2nd score=βtime from transplant to risk evaluation*7+βeGFR*37+βProteinuria*log(0.32)+βDSA MFI*0+βg+ptc*2+βi+t*2+βcg*0+βIFTA*3
The IBox score after therapeutic intervention now projects the patient survival to updated 3, 5 and 7 year-allograft survival probabilities of 85%, 78%, and 68% respectively (see
TABLES:
†Delayed graft function was defined as the use of dialysis in the first postoperative week.
‡Transplant baseline characteristics are donor's age, donor's gender, donor's hypertension, donor's diabetes, recipient's age, recipient's gender, HLA mismatches, retransplantation and anti-HLA DSA at the time of transplantation.
†Sautenet, B., et al. “One-year results of the effects of rituximab on acute antibody-mediated rejection in renal transplantation: RITUX ERAH, a multicentre double-blind randomised placebo-controlled trial.” Transplantation 100.2 (2016): 391-399;
‡Eskandary, Farsad, et al. “A Randomised Trial of Bortezomib in Late Antibody-Mediated Kidney Transplant Rejection.” Journal of the American Society of Nephrology (2017): ASN-2017070818.
†http://www.eurotransplant.org/,
‡http://www.unos.org/
†Premaud A, Filloux M, Gatault P, Thierry A, Buehler M, Munteanu E, et al. An adjustable predictive score of graft survival in kidney transplant patients and the levels of risk linked to de novo donor-specific anti-HLA antibodies. PloS one. 2017; 12(7): e0180236,
indicates data missing or illegible when filed
≥500-3,000
Throughout this application, various references describe the state of the art to which this invention pertains. The disclosures of these references are hereby incorporated by reference into the present disclosure.
1. Mills K T, Xu Y, Zhang W, Bundy J D, Chen C S, Kelly T N, et al. A systematic analysis of worldwide population-based data on the global burden of chronic kidney disease in 2010. Kidney Int. 2015;88(5):950-7.
2. Hill N R, Fatoba S T, Oke J L, Hirst J A, O'Callaghan C A, Lasserson D S, et al. Global Prevalence of Chronic Kidney Disease - A Systematic Review and Meta-Analysis. PLoS One. 2016;11(7):e0158765.
3. Evans R W, Manninen D L, Garrison L P, Jr., Hart L G, Blagg C R, Gutman R A, et al. The quality of life of patients with end-stage renal disease. The New England journal of medicine. 1985;312(9):553-9.
4. Meier-Kriesche H U, Schold J D, Srinivas T R, Kaplan B. Lack of improvement in renal allograft survival despite a marked decrease in acute rejection rates over the most recent era. American journal of transplantation: official journal of the American Society of Transplantation and the American Society of Transplant Surgeons. 2004;4(3):378-83.
5. Coemans M, Susal C, Dohler B, Anglicheau D, Giral M, Bestard O, et al. Analyses of the short- and long-term graft survival after kidney transplantation in Europe between 1986 and 2015. Kidney Int. 2018.
6. Perl J. Kidney transplant failure: failing kidneys, failing care? Clin J Am Soc Nephrol. 2014;9(7):1153-5.
7. Stegall M D, Morris R E, Alloway R R, Mannon R B. Developing New Immunosuppression for the Next Generation of Transplant Recipients: The Path Forward. American journal of transplantation : official journal of the American Society of Transplantation and the American Society of Transplant Surgeons. 2016;16(4):1094-101.
8. Vincenti F, Rostaing L, Grinyo J, Rice K, Steinberg S, Gaite L, et al. Belatacept and Long-Term Outcomes in Kidney Transplantation. The New England journal of medicine. 2016;374(4):333-43.
9. Kaplan B, Schold J, Meier-Kriesche H U. Poor predictive value of serum creatinine for renal allograft loss. American journal of transplantation: official journal of the American Society of Transplantation and the American Society of Transplant Surgeons. 2003;3(12):1560-5.
10. He X, Moore J, Shabir S, Little M A, Cockwell P, Ball S, et al. Comparison of the predictive performance of eGFR formulae for mortality and graft failure in renal transplant recipients. Transplantation. 2009;87(3):384-92.
11. Naesens M, Lerut E, Emonds M P, Herelixka A, Evenepoel P, Claes K, et al. Proteinuria as a Noninvasive Marker for Renal Allograft Histology and Failure: An Observational Cohort Study. J Am Soc Nephrol. 2016;27(1):281-92.
12. Yilmaz S, Tomlanovich S, Mathew T, Taskinen E, Paavonen T, Navarro M, et al. Protocol core needle biopsy and histologic Chronic Allograft Damage Index (CADI) as surrogate end point for long-term graft survival in multicenter studies. J Am Soc Nephrol. 2003;14(3):773-9.
13. Lefaucheur C, Loupy A, Hill G S, Andrade J, Nochy D, Antoine C, et al. Preexisting donor-specific HLA antibodies predict outcome in kidney transplantation. J Am Soc Nephrol. 2010;21(8):1398-406.
14. Moore J, He X, Shabir S, Hanvesakul R, Benavente D, Cockwell P, et al. Development and evaluation of a composite risk score to predict kidney transplant failure. Am J Kidney Dis. 2011;57(5):744-51.
15. Shabir S, Halimi J M, Cherukuri A, Ball S, Ferro C, Lipkin G, et al. Predicting 5-year risk of kidney transplant failure: a prediction instrument using data available at 1 year posttransplantation. Am J Kidney Dis. 2014;63(4):643-51.
16. Gonzales M M, Bentall A, Kremers W K, Stegall M D, Borrows R. Predicting Individual Renal Allograft Outcomes Using Risk Models with 1-Year Surveillance Biopsy and Alloantibody Data. J Am Soc Nephrol. 2016;27(10):3165-74.
17. Schold J D, Kaplan B. The elephant in the room: failings of current clinical endpoints in kidney transplantation. American journal of transplantation: official journal of the American Society of Transplantation and the American Society of Transplant Surgeons. 2010;10(5):1163-6.
18. https://www.eurotransplant.org
19. https://unos.org/
20. Eskandary F, Regele H, Baumann L, Bond G, Kozakowski N, Wahrmann M, et al. A Randomized Trial of Bortezomib in Late Antibody-Mediated Kidney Transplant Rejection. J Am Soc Nephrol. 2018;29(2):591-605.
21. Sautenet B, Blancho G, Buchler M, Morelon E, Toupance O, Barrou B, et al. One-year Results of the Effects of Rituximab on Acute Antibody-Mediated Rejection in Renal Transplantation: RITUX ERAH, a Multicenter Double-blind Randomized Placebo-controlled Trial. Transplantation. 2016;100(2):391-9.
22. Rostaing L, Hertig A, Albano L, Anglicheau D, Durrbach A, Vuiblet V, et al. Fibrosis progression according to epithelial-mesenchymal transition profile: a randomized trial of everolimus versus CsA. American journal of transplantation : official journal of the American Society of Transplantation and the American Society of Transplant Surgeons. 2015;15(5):1303-12.
23. Collins G S, Reitsma J B, Altman D G, Moons K G. Transparent Reporting of a multivariable prediction model for Individual Prognosis or Diagnosis (TRIPOD): the TRIPOD statement. Ann Intern Med. 2015;162(1):55-63.
24. Lamb K E, Lodhi S, Meier-Kriesche H U. Long-term renal allograft survival in the United States: a critical reappraisal. American journal of transplantation: official journal of the American Society of Transplantation and the American Society of Transplant Surgeons. 2011;11(3):450-62.
25. Fine J P, Gray R J. A proportional hazards model for the subdistribution of a competing Risk. J Am Stat Assoc 1999;94:496-509.
26. Efron B. Bootstrap Methods: Another Look at the Jackknife. Ann. Stats. 1979;7(1):1-26.
27. Harrell F E, Jr., Lee K L, Mark D B. Multivariable prognostic models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Stat Med. 1996;15(4):361-87.
28. Pencina M J, D'Agostino R B, Sr., Steyerberg E W. Extensions of net reclassification improvement calculations to measure usefulness of new biomarkers. Stat Med. 2011;30(1):11-21.
29. Uno H, Tian L, Cai T, Kohane I S, Wei U. A unified inference procedure for a class of measures to assess improvement in risk prediction systems with survival data. Stat Med. 2013;32(14):2430-42.
30. Cox D R. Note on Grouping. J Am Stat Assoc. 1957 ;52: 543-547.
31. Rao P S, Schaubel D E, Guidinger M K, Andreoni K A, Wolfe R A, Merion R M, et al. A comprehensive risk quantification score for deceased donor kidneys: the kidney donor risk index. Transplantation. 2009;88(2):231-6.
32. Hosmer D W L S. Applied Logistic Regression. 2nd ed: New York, N.Y.: John Wiley & Sons; 2000.
33. Prasad V, Kim C, Burotto M, Vandross A. The Strength of Association Between Surrogate End Points and Survival in Oncology: A Systematic Review of Trial-Level Meta-analyses. JAMA Intern Med. 2015;175(8):1389-98.
34. Sellares J, de Freitas D G, Mengel M, Reeve J, Einecke G, Sis B, et al. Understanding the causes of kidney transplant failure: the dominant role of antibody-mediated rejection and nonadherence. American journal of transplantation : official journal of the American Society of Transplantation and the American Society of Transplant Surgeons. 2012;12(2):388-99.
35. Premaud A, Filloux M, Gatault P, Thierry A, Buchler M, Munteanu E, et al. An adjustable predictive score of graft survival in kidney transplant patients and the levels of risk linked to de novo donor-specific anti-HLA antibodies. PLoS One. 2017;12(7):e0180236.
| Number | Date | Country | Kind |
|---|---|---|---|
| 19305348.5 | Mar 2019 | EP | regional |
| Filing Document | Filing Date | Country | Kind |
|---|---|---|---|
| PCT/EP2020/058029 | 3/23/2020 | WO |
