Adverse Gastrointestinal Events With Sodium Polystyrene Sulfonate Use in Patients on Maintenance Hemodialysis: An International Cohort Study

Background: There are concerns regarding the gastrointestinal (GI) safety of sodium polystyrene sulfonate (SPS), a medication commonly used in the management of hyperkalemia. Objective: To compare the risk of GI adverse events among users versus non-users of SPS in patients on maintenance hemodialysis. Design: International prospective cohort study. Setting: Seventeen countries (Dialysis Outcomes and Practice Patterns Study [DOPPS] phase 2-6 from 2002 to 2018). Patients: 50 147 adults on maintenance hemodialysis. Measurements: An adverse GI event defined by a GI hospitalization or GI fatality with SPS prescription compared with no SPS prescription. Methods: Overlap propensity score–weighted Cox models. Results: Sodium polystyrene sulfonate prescription was present in 13.4% of patients and ranged from 0.42% (Turkey) to 20.6% (Sweden) with 12.5% use in Canada. A total of 935 (1.9%) adverse GI events (140 [2.1%] with SPS, 795 [1.9%] with no SPS; absolute risk difference 0.2%) occurred. The weighted hazard ratio (HR) of a GI event was not elevated with SPS use compared with non-use (HR = 0.93, 95% confidence interval = 0.83-1.6). The results were consistent when examining fatal GI events and/or GI hospitalization separately. Limitations: Sodium polystyrene sulfonate dose and duration were unknown. Conclusions: Sodium polystyrene sulfonate use in patients on hemodialysis was not associated with a higher risk of an adverse GI event. Our findings suggest that SPS use is safe in an international cohort of maintenance hemodialysis patients.


Introduction
Individuals on maintenance dialysis are at an exceedingly high risk of recurrent hyperkalemia with regional differences in prevalence and severity. 1,2 Of concern, step-wise elevations in potassium are associated with higher all-cause mortality and, as such, potassium levels are carefully monitored and often require additional therapies. 3,4 Dietary potassium restriction, discontinuation of potassium-elevating medications, augmentation in urinary excretion (if applicable), and modifications in dialytic prescription represent viable means of obtaining potassium control. 5,6 One of the most used therapies to reduce or mitigate hyperkalemia are gastrointestinal (GI) potassium binders. Sodium polystyrene sulfonate (SPS), first approved in by the US Food and Drug Administration (FDA) in June 1958, is the most frequently prescribed potassium cation exchange resin globally. 7 Since introduction and widespread use, GI injury, mainly colonic necrosis, has repeatedly been attributed to the use of both oral and rectal SPS-sorbitol coadministration. 8 A large body of basic science, translational, case reports, case series, and epidemiologic studies have linked SPS use with fatal and non-fatal GI injury, both to the upper and lower GI tract. [9][10][11][12][13][14][15][16][17][18][19][20][21][22][23] The risk of GI injury appears high with lower kidney function, concurrent sorbitol administration, and prolonged GI luminal exposure due to constipation. 21 Recently, a large retrospective population-based cohort study from our group including 40 040 older adults similarly documented a nearly 2-fold higher incidence of non-fatal hospitalization or emergency room visit for serious adverse GI events in 20 020 SPS users within 30 days of first prescription compared with 20 020 matched non-users. 20 However, few studies to date have focused solely on patients requiring kidney replacement therapy, specifically maintenance hemodialysis. Hemodialysis patients are exposed, on average, to the highest mean serum potassium, are at high risk of recurrent hyperkalemia, and are commonly constipated placing them at high risk for GI injury. 20,22,24,25 A recent French nationwide dialysis registry study reported no significant increase in emergency room visits or hospitalizations for adverse GI events in 43 771 SPS users. 26 Whether this was region-specific or consistent in a larger, international cohort remains unclear. As such, we set out to examine the association of SPS use and adverse GI events among hemodialysis patients using the international prospective cohort study (Dialysis Outcomes and Practice Patterns Study, DOPPS). We hypothesized that SPS use was associated with an elevated risk of GI injury in patients on maintenance hemodialysis.

Study Design and Population
We conducted a retrospective cohort study of all adults (18 years or older) on hemodialysis captured in DOPPS. The DOPPS is an international dialysis cohort that collects information on patients on maintenance hemodialysis. Dialysis Outcomes and Practice Patterns Study uses a 2-stage random cluster sample where random faculties in a geographic region are selected followed by a random selection of patients within the facilities. 27 Dialysis Outcomes and Practice Patterns Study collects detailed information on individual patients' demographics, comorbid illnesses, medications, laboratory values, dialysis characteristics, hospitalizations, and mortality. Details on the DOPPS study design and methodology can be found elsewhere. 27 We included all individuals enrolled in DOPPS phases 2 to 6 (2002-2018) with known vascular access and where cause of death data was collected. Individuals were GI n'était pas plus élevé avec l'utilisation de SPS (RR = 0,93; IC 95 %: 0,83-1,6). Les résultats étaient cohérents lorsque l'on a examiné séparément les événements gastro-intestinaux (hospitalisation et/ou décès). Limites: La dose et la durée du traitement par SPS étaient inconnues. Conclusion: L'utilisation de SPS chez les patients sous hémodialyse n'a pas été associée à un risque plus élevé d'événements indésirables d'origine gastro-intestinale. Nos résultats suggèrent que l'utilisation du SPS est sans danger dans la cohorte internationale de patients sous hémodialyse d'entretien étudiée.

Keywords
SPS, hyperkalemia, hemodialysis, DOPPS, international, gastrointestinal events   censored at first occurrence of death, study outcome, loss of follow-up, or end of study period (see Supplementary Table 1 for study cohort creation). We excluded individuals with missing vascular access information (n = 1872) and a missing study end date (n = 7). Regional ethics board approval was obtained for this de-identified retrospective study form the Ottawa Hospital Regional Ethics Board (REB).

Exposure
The study exposure was a first SPS prescribed (new user) on periodic (updated every 4 months) medication review since dialysis initiation. Information on dose of SPS use was not available.

Study Outcome
The study outcome was an adverse GI event requiring a hospitalization or a GI fatality. Non-fatal GI events were defined as hospitalization for (1) gastritis/peptic ulcer disease, (2) gastric surgery/resection, or (3) colectomy/colon surgery. Fatal GI events were defined as death due to a direct GI primary cause as deemed by a physician. Fatal GI events were determined using cause of death data and were defined by death by (1) 28,29 were included. The potassium gradient was previously reported to be associated with adverse events in the dialysis population in a dose-dependent manner. 28

Statistical Analysis
We examined differences in baseline characteristics with and without SPS use by standardized differences. Standardized differences describe differences between group means relative to the pooled standard deviation, are less sensitive to large sample sizes than traditional hypothesis testing, and a significant difference is considered to be 10% or greater. 30 We examined the association of SPS use versus non-use and the study outcomes using cause-specific Cox proportional hazards models. 31 Schoenfeld residuals were examined to test the proportionality assumption. We examined SPS exposure as intention to treat and only the first GI hospitalization event was considered. We used overlap propensity score weighting to create balance between the analytical study groups (SPS vs no SPS). 32 A propensity score for SPS exposure was calculated using all the baseline covariates listed in Table 1 and then overlap-weighted analyses were done. Overlap weights assign less weight to observations with extreme propensity scores and more weight to those with propensity scores close to 0.5. 33 This method creates a balanced cohort in terms of the baseline variables and is less influenced by extreme outliers over-dominating often requiring truncation with inverse probability treatment weighting. 34 Post-weighting balance was examined using standardized differences and models were additionally adjusted for variables with imbalance (SD > 0.1), specifically dialysis vintage and the potassium gradient. We conducted all analyses with SAS software, version 9.4 (SAS Institute Inc., Cary, North Carolina). Confidence intervals that did not overlap with 1 or P <.05 were treated as statistically significant.   1.39). The median follow-up time was 1.5 years for total cohort. The median time to a GI hospitalization or a GI fatality was 9.3 and 10.7 months, respectively. These findings were consistent in additional models adjusting for dialysis vintage and the potassium gradient (Table 3). We examined both GI hospitalizations and GI fatality pre-2010 and 2010 onward to examine whether the FDA warning led to a change in outcomes. We found no change in either GI hospitalization (pre-2010: 1.48%, 2010 onward: 1.49%) or GI fatality (pre-2010: 0.63%, 2010 onward: 0.75%) in SPS users.

Discussion
In a large international cohort of individuals on long-term hemodialysis, we found no difference in the risk of an adverse GI event among individuals exposed to SPS use compared with those not exposed to SPS use. No difference was observed when we examined the severity of the GI event separately as fatal or non-fatal. Our findings were consistent in additional models accounting for slight imbalance postweighting and when examining before and after SPS premixing with sorbitol. Our results add to the growing literature on SPS use as a therapeutic agent for the management of hyperkalemia and suggest no detectable GI-related harm in the hemodialysis population. Multiple reports link SPS use to mucosal injury of the GI tract. [8][9][10][11][12][13][14][15][16][17][18][19][20][21][22][23] In a surgical uremic rat model, SPS enema given alone or with sorbitol or mannitol seemed to cause colon necrosis and high mortality rate, whereas 33% sorbitol without SPS did not. 34 In 2013, a systematic review reported that in 17 cases out of the 58 described cases of GI injury, SPS did not contain sorbitol. 22 Furthermore, the colon was the most common site of injury (76%, affecting also the small intestine and the upper GI tract), and transmural necrosis (62%) was the most common histopathologic lesion reported, while mortality was reported in 33% of these cases due to GI injury. In a population-based cohort study of 20 020 SPS users matched to non-users, SPS use was associated with a higher 30-day risk of GI injury requiring an emergency room visit or hospitalizations (0.2% with SPS vs 0.1% without SPS; HR = 1.94; 95% CI, 1.10-3.41). 20 Furthermore, in a Swedish observational study (2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014)(2015)(2016), in patients with chronic kidney disease (CKD), the initiation of SPS was associated with a dose-dependent higher risk of severe GI complications (intestinal ischemia, thrombosis, or ulcer-ation/perforation), as well as the initiation of GI-related medications, compared with non-SPS users. 25 Our results contradict previous findings demonstrating no measurable risk with SPS use and are consistent with a recent French REIN registry study that reported no association with SPS or calcium polystyrene sulfonate (CPS) use and the risk of composite (fatal and non-fatal) GI injury. 25 In models accounting for time-varying exposure, they found no association with SPS or CPS use and GI events of interest (occlusion, perforation, or thrombosis) or other GI events. Previous reports of GI injury with SPS were primarily in non-dialysis populations, or when individuals on dialysis were included, they were in small numbers. This does raise the possibility of a differential risk for those requiring dialysis versus the nondialysis population. Other alternative explanations include differences in study design, handling of the drug exposure (accounting for dose and duration), and possible concurrent drug use that may limit/attenuate the risk (laxatives). The current study expands on these findings by including a larger cohort size with a higher number of events, by the incorporation of important dialysis-related factors such as vascular access, pre-dialysis serum potassium, dialysis time, treatment time, and potassium gradient; and by examining an international dialysis cohort. Our crude rates of GI injury are comparable to those of previous studies, further suggesting appropriate outcome capture.
Our study has limitations. Sodium polystyrene sulfonate exposure was determined by receipt of a prescription and information on adherence, length of use, and dosage were not available. The study exposure was a first SPS prescribed (new user) on periodic (updated every 4 months) medication review since dialysis initiation and our analysis was intention to treat. We did not have information on the route of SPS administration or whether it was co-administered with sorbitol. No pathological, imaging, or autopsy data were available, thus misclassification is possible. Our focus on GI composite events, as opposed to individual types of GI pathology, may partially mitigate this. Despite accounting for many variables in our weighted models, residual confounding may remain a possibility. Post-weighting, a few variables did remain imbalanced (related to potassium and dialysis vintage) that required further adjustment in our analytic models. Nevertheless, our study is the largest, international study capturing physician deemed cause of death. Our findings of no higher risk with SPS use were consistent when we examined different outcomes (fatal or any GI event).

Conclusions
In a large international hemodialysis prospective cohort study, we found SPS use was not associated with a higher risk of a GI adverse event, and this was consistent in additional analyses. Further research to delineate the GI injury risk by dose-dependent SPS exposure and long-term use in the maintenance hemodialysis population is warranted.

Ethics Approval and Consent to Participate
Regional ethics board approval was obtained for this de-identified retrospective study form the Ottawa Hospital Regional Ethics Board (REB).

Consent for Publication
Not applicable.

Availability of Data and Materials
DOPPS data is not freely available but access may be granted to investigators and collaborators via a data request.