Paper Details

ABSTRACT

Background:

Cutaneous T-cell Lymphomas (CTCLs) are rare lymphomas with particularly debilitating symptom burden. Given the impact CTCL and its treatment have on patients’ quality of life (QoL), the patient experience is essential to incorporate into clinical decision-making and can be measured by patient reported outcomes (PROs). Though the importance of PROs in the study of CTCL is growing, how PROs change over a patient’s treatment course and the extent to which PROs are linked with prognosis and survival have not been well characterized. To understand how baseline disease and patient characteristics influence PROs and how baseline PROs may predict overall disease trajectory, we analyzed data from patients with CTCL from a prospective multicenter cohort with longitudinal PROs.

Methods:

The LEO cohort study and Molecular Epidemiology Resource enrolled patients with CTCL (2002-2020) and collected Functional Assessment of Cancer-general (FACT-G) and FACT-lymphoma (FACT-Lym) at baseline and at months 12, 24, 36, and 60 after diagnosis. Minimally important differences (MIDs) were defined as 3 for FACT-G and 6 for FACT-Lym. The Wilcoxon rank-sum and Kruskal–Wallis tests were used for comparing PRO scores by baseline characteristics. Linear mixed-effects models (LMM) with patient-level random intercepts were used to evaluate longitudinal changes in PRO scores and the association of relapse status with PRO scores over time. Association of PRO scores with overall survival (OS) and event-free survival (EFS) was assessed with multivariable Cox proportional hazards regression models, adjusting for sex, race, ethnicity, age, stage, subtype, and initial therapy. Kaplan–Meier survival curves were used to compare survival between groups (high vs. low PRO scores split by the sample median) and the log-rank test to assess statistical significance.

Results:

231 patients with CTCL completed at least a baseline PRO measure (68 were pre-treatment, 186 completed FACT-G, 114 completed FACT-Lym). By CTCL subtype, 156 patients had mycosis fungoides, 32 had Sezary syndrome, 47 had primary cutaneous anaplastic large cell lymphoma (pcALCL), 22 had subcutaneous panniculitis-like T-cell lymphoma (SPTCL), and 24 were other. No differences were observed in baseline PRO scores based on sex, race, ethnicity, age, or education level. Patients with baseline ECOG <2 had better baseline PRO scores (ECOG <2 FACT-G 86.0, FACT-Lym 135.4) compared to patients with a baseline ECOG ≥2 (FACT-G 62.3 p<.001, FACT-Lym 89.4 p=.002). Patients with SPTCL had the worst baseline FACT-G scores and patients with pcALCL had the best baseline FACT-G scores (72.4 and 89.7 respectively, p=.04). For patients with available paired data (n=51), disease relapse was significantly associated with worse FACT-G scores in the LMM (β = −2.9, 95% CI: -5.6 to -0.32, p=0.03) compared to non-relapse assessments, though the estimated difference did not reach the MID. Using the 3-point MID threshold for FACT-G, after relapse 26/51 (51%) patients had worse PRO scores, 13/51 (25%) had no meaningful change, and 12/51 (24%) patients had improved PRO scores. In a univariate analysis, EFS was longer in patients with high/better baseline FACT-G/FACT-Lym scores compared to low/worse baseline FACT-G/FACT-Lym scores (low FACT-G mEFS=21.0 vs. high FACT-G mEFS=59.3, p=.008, low FACT-Lym mEFS=21.0 vs. high FACT-Lym mEFS=59.6, p=.02). However in the multivariable model, FACT-G and FACT-Lym were not independent predictors of EFS (HR 1.00/1.01, p>0.9/p=.4). In a univariate analysis, OS was significantly longer in patients with high/better baseline FACT-G scores compared to low/worse baseline FACT-G scores (low FACT-G mOS=81.9 months, high FACT-G mOS=not reached, p=.008). However in the multivariable model, FACT-G was not an independent predictor of OS (HR .98, p=.06).

Conclusion:

Although findings were limited by PRO collections at pre-specified time points, rather than aligned with events, QoL declines significantly for the majority of patients following a relapse event, indicating that QoL changes dynamically with disease status and should possibly be incorporated into response to therapy assessment. FACT-G/FACT-Lym alone are not prognostic for survival, suggesting the need for more disease specific measures that are prognostic as well as aligned with change in disease status to be used as clinical trial endpoints. To our knowledge, this is one of very few studies evaluating serial PROs in a CTCL cohort.