Aim

To summarize the details and findings of a retrospective analysis conducted by Brodin-Sartorius et al.²²

Background

The manufacturer provided a retrospective cohort study by Brodin-Sartorius et al., of 86 nephropathic cystinosis patients (aged ≥ 15 years old) diagnosed in France between 1961 and 1995, as the clinical basis for the time to event data used to populate the pharmacoeconomic model.²² Patient data were collected retrospectively from two hospitals (l’hôpital Edouard Herriot [Lyon] and l’hôpital Necker-Enfants malades [Paris]) by asking pediatric and adult nephrologists about their patients.

Results

Immediate-release cysteamine therapy was administered to 75 patients (87%) starting at a mean age of 9.9 years (median 4.3 years; range 0.9 years to 38.6 years) for a mean duration of 17.4 years (median 18.4 years; range 0.9 years to 28.4 years). Patients were divided into three subgroups: a group that started treatment before age 5 years (n = 40 patients); a group that started treatment at or after age 5 years (n = 8 patients); and a group that was untreated prior to the development of end-stage renal disease (ESRD) (n = 38 patients). Twenty-seven patients in the latter subgroup started cysteamine treatment at a mean age of 23 years (median 22.1 years; range 18.9 years to 24.8 years); 11 patients in this group never received cysteamine.

Physician-reported adherence to therapy was reported as “good” for 26 patients (34.6%) and “quite good” for 31 patients (41.3%). A total of 21 patients (28%) had “extended periods without treatment,” and seven patients were missing adherence information. Mean white blood cell (WBC) cystine level recorded in 78 patients during the overall follow-up was less than 2 nmol half cystine/mg protein for 22 patients (28.2%), between 2 and 3 nmol half cystine/mg protein for 27 patients (34.6%), and greater than 3 nmol half cystine/mg protein for 29 patients (37.2%).

The outcomes were the occurrence of survival, renal complications (primarily ESRD), extrarenal complications (hypothyroidism, neuromuscular disorder, or diabetes), and socioprofessional outcomes.

Twenty-four patients in the cohort died (27.9%): of those, seven (29.2%) never received treatment, two (8.3%) started cysteamine before 5 years of age, and 15 (62.5%) who started cysteamine after 5 years of age. Seven patients who were never treated (7/11; 64%) died during the study. Survival was significantly improved in the patients treated before the age of 5 years when compared with those not treated (log-rank P = 0.03). Patients who started cysteamine after 5 years of age also had significantly improved survival compared with untreated patients (log-rank P < 0.05).

Cysteamine therapy initiated before the age of 5 years significantly decreased the incidence of ESRD compared with those who were untreated and those who started after 5 years. The mean age at ESRD onset was 13.4 ± 4.8 years (median 12.2 years) in the group that started treatment before 5 years of age, and 9.6 ± 2.6 years (median 9.5 years) in the group that started treatment after 5 years of age (P < 0.05). Kaplan–Meier survival curves showed that starting cysteamine therapy before the age of 5 years significantly delayed the ESRD onset (log-rank P <0.0001) relative to those who started after age 5 years. Patients rated by their physicians as having good adherence had significantly delayed ESRD in comparison with patients with lower adherence or those not treated (log-rank P < 0.0001).

A significant delay in the occurrence of neuromuscular disorders, hypothyroidism, and diabetes in nephropathic cystinosis patients treated with cysteamine before the age of 5 years compared with untreated patients (log-rank, P < 0.001), was also reported. A statistically significant delay in diabetes and hypothyroidism was reported for patients who started treatment after the age of 5 years compared with the untreated group (P < 0.001).

Limitations

The identification of patients for inclusion in the cohort (i.e., patients with a diagnosis of nephropathic cystinosis) appeared to be appropriate, and was based on the presence of corneal cystine deposits and/or an elevated WBC cystine level of greater than 3 nmol half cystine/mg protein. However, it is not clear if the cohort comprised all patients aged greater than and equal to 15 years of age with a diagnosis of nephropathic cystinosis between 1961 and 1995 seen at the specified two hospitals in France, or if there were any exclusions. Also, the article provides minimal information on the exact sources of data for the study, other than stating that “data were collected from historically referent hospitals (Lyon–Herriot and Paris–Necker Hospital),” and “pediatric and adult nephrologists were asked about their patients.”²² Therefore it was not possible to assess the validity of the data sources.

It is uncertain whether the seemingly small number of patients in the untreated group (n = 11) and within each of the three subgroups (e.g., a total of eight patients in the group that initiated treatment at greater than and equal to 5 years of age) was truly sufficient for the statistical comparisons. Statistically significant comparisons with small P values were reported for many of the comparisons, suggesting that statistically there were enough patients in the study groups to detect a difference beyond chance. It was also reported that adjustments for multiplicity were performed: for example, pairwise comparisons were adjusted with the Hochberg approach and Kaplan–Meier analyses with overall log-rank testing and Bonferroni method for two-by-two comparisons were used for time to event analyses. However, no additional information was provided as to whether the reported P values were the adjusted values, and given the number of comparisons made there is a high probability of type I error inflation if the comparisons were not adequately adjusted for multiple comparisons. Moreover, most of the comparisons were based on hypothesis tests with P values but standard errors and confidence intervals were not reported, making it difficult to evaluate the precision and the clinical relevance of the statistical findings.

The small numbers likely also limited the investigators’ ability to examine and statistically adjust for confounders and/or effect modifiers. It appeared that all analyses were unadjusted, producing crude estimates. The investigators did test for differences between groups with respect to baseline characteristics, and identified statistically significant differences between the subgroups for age, number of follow-up years, and number of nephropathic cystinosis-related complications. There are limitations to conducting hypothesis tests on baseline characteristics (e.g., not fulfilling certain underlying assumptions for the tests); however, the fact that no evaluation of the potential impact of differences in baseline characteristics (e.g., effect modifiers) was undertaken, means that there is some possibility that the results could be explained by other factors.

As mentioned, the sources for information were vaguely described in the article. Therefore a key point in the analysis, proper classification of patients based on exposure, is uncertain. Hence, it is difficult to state with certainty, based on the information reported, that exposure misclassification did not occur. With respect to the outcomes, the description of outcome assessment suggested that outcomes were assessed in a clinically appropriate manner. Given that the outcomes are based on reasonably objective measures, it is unlikely that misclassification of outcome occurred.

Survival bias, as acknowledged by the authors, is a possibility in this study. The time frame for the analysis spanned 34 years; therefore there is a reasonable probability that there were cohort period differences. The earliest diagnosed patients in the 1960s would have had different disease management and supportive care than those diagnosed closer to 1995 (e.g., improvement in supportive treatment for renal and extrarenal complications), which could potentially have contributed to higher survival rates for patients diagnosed later in the study.

A key component of the analysis was an assessment of patient adherence to cysteamine treatment. Adherence was measured in the study based on physician questioning of patients, coded in loosely defined categories: “good,” “quite good,” or “extended period without treatment.” No information was provided on whether this was validated or if coding was verified. Therefore, adherence may have been misclassified.

Conclusions

Brodin-Sartorius et al. present a long-term study on the impact of cysteamine therapy on nephropathic cystinosis complications. The data suggest that cysteamine delays death and other complications of nephropathic cystinosis but does not prevent them. There were some key limitations to the study, as mentioned. However, the results of the study are generally consistent with clinical practice, as described by the clinical expert involved in the CDR review. In particular, earlier initiation of cystine-depleting treatment appears to have an impact on delaying death and complications associated with nephropathic cystinosis.