Wen Yao Mak(1,2), Aole Zheng(1), Chee Kin Yoon(2), Loke Meng Ong(2), Xiaoqiang Xiang(1), Xiao Zhu(1,)
(1)Department of Clinical Pharmacy and Pharmacy Administration, School of Pharmacy, Fudan University, Shanghai, China (2)Clinical Research Centre (Penang General Hospital), Institute for Clinical Research, National Institute of Health, Malaysia
BACKGROUND
Peritonitis is a major complication of peritoneal dialysis (PD)[1]. Despite recent progress in PD product safety, studies still showed that a typical patient would experience one peritonitis every 9-12 month[2]. Conventional analysis of peritonitis hazards typically depended on non- or semi-parametric approach to identify risks factors[3-5], but these tools are inadequate to capture the complex relationships of real-world data (RWD), particularly between nonlinear or time-varying factors that are often encountered in clinical practice. Patient registries (e.g. Malaysian Renal Registry) are valuable resources for public health and care management planning but conventional survival analysis have methodological limitations to fully derive insights from these datasets. Conversely, parametric repeated time-to-event (RTTE) analysis is favored as it considers the event time itself as a variable and could handle censored data. The pharmacometrics (PMX) RTTE approach could also quantify interindividual variabilities (IIV), rendering it well suited to uncover novel predictors of repeated peritonitis over the entire PD timespan.
OBJECTIVE
The study aimed:
- To determine the predictors of repeated peritonitis in PD patients over the duration of dialysis
- To evaluate the performance of RTTE methods to generate novel insights based on RWD in comparison to conventional Kaplan-Meier (KP) and Cox proportional hazard (PH) analysis
METHODS
Patient data
The dataset consisted of 10,084 end-stage renal failure patients from 35 Malaysian public hospitals who started PD from the inception of the renal registry (2008) until December 31, 2018 (censor date). All patients were tracked from the date of first dialysis until one of these outcomes were met: death, transfer to hemodialysis, renal transplant, or censored. All incidents of peritonitis were recorded (precise to the day of diagnosis). Baseline demographic information and relevant time-varying covariates such as laboratory test results and dialysis sufficiency were extracted for analysis. Exploratory KP and Cox PH analyses were performed to identify the predictors of first peritonitis event.
Model development
Different baseline hazard parameterizations were tested, including the exponential, Weibull, Gompertz, log-normal and log-logistic distributions. Base hazard model selection was based on AIC. Interindividual variability (IIV) was included on hazard as a measure of the change in peritonitis frequency. A stepwise covariate model (SCM) building procedure was performed for covariates searching (forward selection: p<0.05, backward elimination: p<0.001). For missing time-varying covariates, last observation carry forward was applied; for baseline covariates, the median was imputed. Both baseline and time-varying covariates were included into the model exponentially. The final model was evaluated using Kaplan-Meier visual predictive checks (VPCs) for time-to-first and subsequent events.
RESULTS
The mean (SD) age of the cohort was 57 (15.1) years, with 48.6% patients being female. A total of 6,549 cases of peritonitis were recorded, with a median (interquartile range) follow-up period of 4.32 (2.35-6.96) years. Exploratory KP and Cox PH identified prior exit site infection (ESI), heart failure, and psychological disorder as predictors of first peritonitis.
The final RTTE model included a Gompertz distribution for baseline hazard, and was influenced by prior ESI, time-varying hemoglobin (HB) and albumin (ALB). Kaplan-Meier VPC indicated the observed time-to-first and subsequent peritonitis data was sufficiently described by the final RTTE model. Model parameters (values, [%RSE]) were precisely estimated, and predicted an increasing hazard of peritonitis over time (scale, λ=0.178 [3%]; shape, β=0.004 [28%]), which was aggravated by the presence of prior ESI (0.789 [2%]), low albumin (-0.0398 [4%]) and hemoglobin (-0.0095 [30%]) levels (biomarkers of worsening renal functions). Inclusion of covariates reduced the IIV from 116% [5%] of the base model to 88.8% [9%]. Compared to KP and Cox PH analyses, the RTTE model uncovered novel time-varying influences of ALB and HB on peritonitis hazards.
CONCLUSION
In this study, prior ESI, ALB and HB were identified as important contributors to peritonitis risks over time. The PMX method complemented existing analytical toolbox in understanding complex real-world clinical data, and could provide much added value in guiding clinical practice.
References:
[1] Y. Cho and D. W. Johnson, ‘Peritoneal Dialysis–Related Peritonitis: Towards Improving Evidence, Practices, and Outcomes’, Am. J. Kidney Dis., vol. 64, no. 2, pp. 278–289, Aug. 2014, doi: 10.1053/j.ajkd.2014.02.025.
[2] R. Maiorca et al., ‘A six-year comparison of patient and technique survivals in CAPD and HD’, Kidney Int., vol. 34, no. 4, pp. 518–524, Oct. 1988, doi: 10.1038/ki.1988.212.
[3] Y. He et al., ‘Decreased Peripheral Blood Lymphocyte Count Predicts Poor Treatment Response in Peritoneal Dialysis-Associated Peritonitis’, J. Inflamm. Res., vol. 16, pp. 5327–5338, Nov. 2023, doi: 10.2147/JIR.S438674.
[4] Z. Zhao, Q. Yan, D. Li, J. Duan, D. Liu, and Z. Liu, ‘Relationship between serum iPTH and peritonitis episodes in patients undergoing continuous ambulatory peritoneal dialysis’, Front. Endocrinol., vol. 14, Mar. 2023, doi: 10.3389/fendo.2023.1081543.
[5] S. Ljungman et al., ‘Factors associated with time to first dialysis-associated peritonitis episode: Data from the Peritonitis Prevention Study (PEPS)’, Perit. Dial. Int. J. Int. Soc. Perit. Dial., vol. 43, no. 3, pp. 241–251, May 2023, doi: 10.1177/08968608231161179.
Reference: PAGE 32 (2024) Abstr 11138 [www.page-meeting.org/?abstract=11138]
Poster: Real-world data (RWD) in pharmacometrics