Article Text
Abstract
Background/Aims A series at a single clinical centre recently demonstrated an association between the interstitial cystitis drug pentosan polysulfate sodium (PPS) and a vision-threatening pigmentary maculopathy. The aim of this study was to determine if an association exists between PPS use and macular disease in a large national cohort.
Methods A retrospective, matched cohort study using data from a large US medical claims database from 2002 to 2016 was performed. A total of 3012 and 1604 PPS users were compared with 15 060 and 8017 matched controls at 5 and 7 years, respectively. The primary outcome measures included (1) any new diagnosis of a hereditary or secondary pigmentary maculopathy (atypical maculopathy outcome), and (2) any new diagnosis of dry age-related macular degeneration (AMD) or drusen in addition to the aforementioned diagnoses (atypical maculopathy+AMD outcome).
Results At the 5-year and 7-year follow-up, 9 (0.3%) and 10 (0.6%) PPS patients progressed to the atypical maculopathy outcome compared with 32 (0.2%) and 25 (0.3%) control patients, respectively. 103 (3.4%) and 87 (5.4%) PPS patients developed the atypical maculopathy+AMD outcome compared with 440 (2.9%) and 328 (4.1%) control patients at 5 and 7 years, respectively. At 5 years, multivariate analysis showed no significant association (p>0.13). At 7 years, PPS users had significantly increased odds of having the atypical maculopathy+AMD outcome (OR=1.41, 95% CI 1.09 to 1.83, p=0.009).
Conclusions PPS exposure was associated with a new diagnosis of macular disease at the 7-year follow-up in a large national cohort.
- macula
- drugs
Statistics from Altmetric.com
Introduction
Pentosan polysulfate sodium (PPS; Elmiron (Janssen Pharmaceuticals, Titusville, New Jersey, USA)) is a semisynthetic macromolecule approved for managing the pain or discomfort associated with interstitial cystitis (IC).1 Also referred to as bladder pain syndrome, IC is a chronic disease complex characterised by urinary urgency, nocturia, urinary frequency, and bladder or pelvic pain that affects over 1 million US adults.2 PPS is the only Food and Drug Administration (FDA)-approved oral therapy for IC and has been widely prescribed since its approval in 1996.3 The American Urological Association Interstitial Cystitis Treatment Guidelines recommend PPS as a second-line therapy after non-pharmacological interventions such as patient education, stress management and behavioural modification.4
We recently described a unique pigmentary maculopathy associated with chronic PPS exposure in a series of patients seen at a single clinical centre.5 6 These patients reported disabling symptoms of prolonged dark adaption and difficulty reading. Although fundus examination findings were subtle, modern retinal imaging techniques revealed striking and characteristic fundus abnormalities. Most patients carried an initial diagnosis of either age-related macular degeneration (AMD) or pattern dystrophy prior to being reclassified as having a PPS-associated maculopathy.
These findings regarding a possible toxic maculopathy represent a major patient safety issue, and further study is warranted to explore this association. The aim of this study was to evaluate the association between PPS use and a new diagnosis of a macular disorder using an administrative claims database from a large US insurer.
Methods
Dataset
A retrospective, matched cohort study using data from the Clinformatics Data Mart Database (OptumInsight, Eden Prairie, Minneapolis, USA) was performed. This database was created from a large national US insurer’s deidentified administrative medical claims and includes all outpatient medical claims, associated diagnoses, pharmaceutical prescriptions filled and demographic data for all beneficiaries during their enrolment in the insurance plan. Since this study involves deidentified data, it was declared exempt from review by the University of Pennsylvania’s Institutional Review Board.
Cohorts
An exposure cohort was created from all patients who filled a prescription for PPS from 1 January 2002 to 31 December 2016. The index date for each patient was set to the date of the first prescription filled. Patients were excluded for not having at least 2 years of time in the insurance plan prior to the index date or for having any previous diagnosis of a retinal toxicity, hereditary retinal degeneration, AMD or drusen. An unexposed cohort was then created by matching 5 to 1 for every exposed patient and was matched on age (±5 years), sex, race and insurance start date (±3 months). The index date of the exposed was then assigned to the unexposed control and the above exclusion was applied. If an exposed control was excluded, another match was found that met criteria to maintain 5:1 matching. The analysis also required that all exposed patients and unexposed controls have at least 5 years of follow-up after the index date. A second analysis required that patients have at least 7 years of follow-up. Due to beneficiary attrition, we were unable to reliably assess time points longer than 7 years. (See online supplemental table 1 for all International Classification of Diseases (ICD)-9/ICD-10 codes used within this study.)
Supplemental material
Outcomes and covariates of interest
The primary outcome was addressed in two ways: (1) any new diagnosis of a hereditary or secondary pigmentary retinopathy (atypical maculopathy outcome) and (2) any new diagnosis of dry AMD or drusen in addition to the aforementioned diagnoses (atypical maculopathy+AMD outcome). These outcomes were based on the resemblance of the newly described maculopathy to other retinal diseases. Covariates of interest were age, hypertension, diabetes mellitus, transient ischaemic attack or stroke, year of index date, peripheral vascular disease, malignancy, atrial fibrillation/flutter, congestive heart failure, previous myocardial infarction, arrhythmia, yearly income, education level and geographical region.
Statistical analysis
Demographic data and comorbid conditions were collected at the time of the index date. Mean and SD were used to summarise continuous variables, while frequencies and percentages were used for categorical variables. Logistic regression was performed to evaluate the odds of developing a macular disease at 5 and 7 years. Candidate variables for final multivariate modelling were chosen for association (p<0.2) in univariate modelling. Additional sensitivity analyses were run to test assumptions within our primary model. Since PPS is primarily given to patients with IC, we mandated everyone in both the exposed and the unexposed cohorts to have a cystitis diagnosis to be certain macular changes were associated with the medication and not the disease process being treated. Next, since the pigmentary maculopathy is thought to be secondary to chronic exposure of PPS, we assessed prescription days of coverage at 5 and 7 years for the PPS patients. Statistical analysis was performed using SAS (V.9.4; SAS Institute).
Results
A total of 3012 patients filled a prescription for PPS and met the inclusion criteria at the 5-year follow-up (figure 1). Further, 15 060 controls were matched to these exposed patients. Due to the matching, there was no difference in mean age (52.3 (SD 15.7) years; p=0.98), gender (both 84% female) or racial composition (71% white, 14% Hispanic, 8% black, 2% Asian, 5% unknown). Table 1 shows all baseline covariates.
At the 5-year follow-up, 9 (0.3%) PPS users compared with 32 (0.2%) control patients progressed to the atypical maculopathy outcome. At the same time point, 103 (3.4%) PPS users compared with 440 (2.9%) control patients developed the more inclusive atypical maculopathy+AMD outcome. Multivariate logistic regression analysis showed no significant association for the atypical maculopathy outcome (OR=1.35, 95% CI 0.64 to 2.83, p=0.43) or the atypical maculopathy+AMD outcome (OR=1.20, 95% CI 0.95 to 1.51, p=0.13) at the 5-year follow-up (table 2).
For the 7-year follow-up, a total of 1604 PPS users met the inclusion criteria, and 8017 controls were matched to the exposed patients (figure 1). Comparison of the baseline characteristics between the two cohorts revealed no difference in mean age (52.8 (SD 15.3) years; p=1.00), gender (83% female) or race (70% white, 14% Hispanic, 8% black, 2% Asian, 5% unknown). Table 3 shows all baseline covariates used in the comparison of baseline characteristics.
At the 7-year follow-up, 10 (0.6%) PPS users compared with 25 (0.3%) control patients developed the atypical maculopathy outcome, while 87 (5.4%) PPS users compared with 328 (4.1%) control patients progressed to the atypical maculopathy+AMD outcome. Univariate logistic regression analysis showed increased odds of developing the atypical maculopathy outcome (OR=2.01, 95% CI 0.96 to 4.18, p=0.06), although the association did not reach significance. However, for the atypical maculopathy+AMD outcome, univariate analysis did show a significantly increased OR for PPS users (OR=1.34, 95% CI 1.06 to 1.71, p=0.02). Multivariate analysis reproduced similar findings, revealing no significant association for PPS use and the atypical maculopathy outcome (OR=1.87, 95% CI 0.87 to 3.99, p=0.11) and a significant association between PPS use and the atypical maculopathy+AMD outcome (OR=1.41, 95% CI 1.09 to 1.83, p=0.009) (table 4).
Sensitivity analyses that required a diagnosis of cystitis were also performed at 5 and 7 years. A total of 12 096 patients were included in the 5-year analysis, of which 2016 patients were exposed to PPS and 10 080 patients were not exposed. Multivariate logistic regression demonstrated significantly increased odds of PPS users having the atypical maculopathy outcome (OR=2.91, 95% CI 1.15 to 7.36, p=0.02) at the 5-year follow-up. However, the atypical maculopathy+AMD outcome revealed no significant association (OR=1.00, 95% CI 0.76 to 1.30, p=0.97) at this time point. Sensitivity analyses at the 7-year follow-up included a total of 6408 patients (1070 in the exposed cohort and 5338 in the unexposed cohort) and similarly did not find a significant association between PPS use and development of the atypical maculopathy outcome (OR=1.46, 95% CI 0.66 to 3.24, p=0.35) or the atypical maculopathy+AMD outcome (OR=0.94, 95% CI 0.69 to 1.28, p=0.69). In addition, consistency of PPS use was evaluated. After 5 years of initiating PPS, the average user had 312.2 days of prescription coverage. At 7 years, the PPS prescription coverage increased to 391.1 days of use.
Discussion
In this study of medical claims in a large national cohort, we found a significant association between PPS exposure and a new diagnosis of macular disease at 7 years. These results corroborate findings of the initial case series that identified an association in six patients at a single clinic.5 Given the broad use of PPS for IC, these findings suggest that thousands of additional patients may be at risk for a vision-threatening drug-associated maculopathy.
When designing our outcomes, we tailored the list of retinal disease codes to best represent the spectrum of phenotypes of this newly described condition. In the initial series at our tertiary referral centre, a majority of patients presented with a diagnosis of pattern dystrophy or AMD, typically given by a referring retina specialist. These diagnoses are a natural fit for this previously unknown condition, given the typical retinal manifestations of macular pigment clumping, atrophy and subretinal yellow deposits. Thus, we included codes for a range of retinal dystrophies, acquired pigmentary retinopathies, macular toxicity and AMD (online supplementary table 1).
Chronicity of PPS use likely plays a role in the development of this maculopathy. Although the initial series on PPS-associated maculopathy reported exposure durations exceeding 10 years in all patients, we have recently identified numerous patients in our clinical practice that have manifested the characteristic maculopathy with shorter exposure durations.7 In the present study, we were unable to perform the analysis at 10 years due to patient attrition. However, our results suggest that differences may present with as few as 7 years of exposure. Additionally, although not significant, there were increased odds of having macular disease at 5 years. We predict that a longer follow-up interval would demonstrate even greater odds of developing a macular disease in the PPS-exposed group. These results have implications for timing of visits in drug safety surveillance programmes.
It is unusual for a potential drug toxicity to manifest decades after initial FDA approval. Several previous long-term studies of PPS did not identify any notable ocular safety signals.1 8 9 In the longest open-label study to our knowledge, with some patients receiving extended treatment for 90 months, no ocular adverse events were identified.
We believe that several factors have led to the delayed recognition of this entity. First, patients with IC are often suffering from multiple medical comorbidities and under treatment with numerous medications, making it difficult to parse out a drug-disease association. Second, many patients are likely just now reaching the exposure threshold to manifest this condition. Third, the fundus findings in this condition are subtle and resemble other well-known maculopathies such as AMD and pattern dystrophy. The pattern dystrophy diagnosis, in particular, is increasingly applied to an ill-defined collection of pigmentary maculopathies.10 Given the relatively benign course and lack of treatment for pattern dystrophies, patients with this diagnosis do not typically undergo extensive diagnostic evaluations.
The mechanisms of PPS-associated maculopathy remain unclear. PPS is a semisynthetic macromolecule with a structure similar to glycosaminoglycans. Retinal imaging suggests a primary abnormality within the retinal pigment epithelium (RPE) or at the RPE-photoreceptor interface.4 It is possible that the drug or one of its many metabolites causes toxicity to the RPE or disturbs the integrity of the interphotoreceptor matrix, which is comprised largely of glycosaminoglycans. Further investigation into the pathogenesis of disease is warranted. Given the nature of macular changes noted, this work may reveal unanticipated insights into retinal biology and mechanisms of diseases such as AMD.
Several considerations need to be made when assessing the results of this study. First, PPS users only averaged 312 and 391 days (roughly 10 and 13 months) of prescription coverage, despite 5 and 7 years of observation. PPS is only available as a branded medication in the USA and initial therapy recommendation is three pills per day for 90 days11 at a cost of roughly $970/100 pills.12 Although by definition the patients in this study had insurance, it has been shown that increased copays reduce adherence to prescription medications.13 If as suggested by the case series, the exposure needs to be chronic for the maculopathy to present, then it is likely we had relatively few patients reach the threshold for maculopathy to occur. Next, due to the observational nature of the dataset, many patients likely never received an eye examination during their time in the plan, making it harder to detect a signal within this database (and any association found would likely underestimate the true risk). Next, we identified patients who filled a prescription of PPS, yet we were unable to confirm how much of that prescription was actually ingested as prescribed. Reduced usage in patients thought to take a medication would cause a bias to the null. Despite all of these aspects of the claims database used in this study increasing the likelihood of having a null study, we still found evidence that PPS use is associated with a maculopathy.
Additional potential limitations also need to be considered. Any study of this type has a risk of indication bias, where the reason for giving the drug is actually the cause of the association seen, and not the drug itself. We performed a sensitivity analysis that only included cystitis patients. While the confirmation of the atypical maculopathy association was reassuring for the PPS association at 5 years, we were not able to replicate the findings at 7 years or in the atypical maculopathy+AMD analysis. Possible reasons for this include a lowered powered analysis due to smaller sample size or an association for patients with cystitis (possibly through other associated comorbidities) to have an increased risk of AMD. Lastly, due to the nature of the dataset, we were unable to verify specific diagnoses with chart review, nor were we able to control for smoking.
In conclusion, our study of medical claims data from a large US national cohort demonstrates a statistically significant association between PPS use and the development of macular disease at 7 years. The results of this study are far-reaching, as many thousands of patients have received PPS since its FDA approval in 1996.14 Further investigations will explore pathogenesis of disease, clarify the spectrum of disease phenotypes and inform screening guidelines.
Acknowledgments
This work was presented at the Annual Meeting of the Association for Research in Vision and Ophthalmology (ARVO), Vancouver, BC, in May 2019.
Footnotes
Contributors NJ: study design and conception, data analysis and interpretation, writing manuscript, manuscript revision, final approval and agreement to be accountable. AL: data analysis and interpretation, writing manuscript, manuscript revision, final approval and agreement to be accountable. YY: data acquisition, data analysis and interpretation, manuscript revision, final approval and agreement to be accountable. BLVB: study design and conception, data acquisition, data analysis and interpretation, writing manuscript, manuscript revision, final approval and agreement to be accountable.
Funding Foundation Fighting Blindness (CD-C-0918-0748-EEC) (NJ); NIH Core Grant P30 EY006360 (Emory Eye Center); National Institutes of Health K23 Award (1K23EY025729-01) (BLVB) and University of Pennsylvania Core Grant for Vision Research (2P30EY001583) (BLVB, YY). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. Additional funding was provided by Research to Prevent Blindness and the Paul and Evanina Mackall Foundation. Funding from each of the above sources was received in the form of block research grants to the Scheie Eye Institute. The sponsors or funding organisations had no role in the design or conduct of this research.
Competing interests None declared.
Patient consent for publication Not required.
Ethics approval Ethics approval was not required due to the deidentified nature of the dataset. It met the eligibility criteria for Institutional Review Board review exemption (Protocol No. 819924).
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement Data may be obtained from a third party and are not publicly available.