Enrolment in clinical trials is a key objective to improve the development of new treatment strategies, but due to stringent inclusion/exclusion criteria, patient participation is less than 10%.1 In addition to interventional studies, there is an important development of real-world data (RWD) studies.2,3 One objective of prospective real-life multicentric cohorts is to collect data closer to the general population.4,5 REALYSA study is a prospective real-life multicentric cohort set up in France to study the prognostic value of epidemiological, clinical and biological factors (NCT03869619).6 Adult patients with newly diagnosed diffuse large B-cell (DLBCL), follicular (FL), marginal zone (MZL), mantle cell (MCL), Burkitt, Hodgkin (HL), mature T-cell lymphomas were invited to participate by their haematologist. Exclusion criteria were having already received anti-lymphoma treatment (except prephase), posttransplantation lymphoproliferative disorder and documented HIV infection.
Determinants that influence patient participation in RWD programs are not well known. We aimed to investigate the variables associated with inclusion in REALYSA among these seven newly diagnosed lymphoma subtypes who were managed at the Haematology Department, Lyon Sud Hospital, France between 1st November 2018 and 30th November 2019 using REALYSA and institutional databases.
We modelled the probability of being enrolled in REALYSA using generalized additive models (GAMs) (Supplementary Methods).7 For categorical covariates, we reported odds ratio (OR) of REALYSA enrolment. Age, distance between home and hospital, albumin, LDH at diagnosis were studied as continuous covariates. We exploratory replicated GAM analyses and estimated event-free survival (EFS) for major lymphoma subgroups (DLBCL, FL, HL) according to REALYSA participation.
Among the 278 patients fulfilling the eligibility criteria of REALYSA recruited over one year in our Haematology Department, 57% were male, and the median age was 64 years (18-96). The lymphoma subtypes were as follows: 104 DLBCL (37%), 56 FL (20%), 46 HL (17%), 35 MZL (13%), 19 T-cell lymphoma (7%), 16 MCL (6%), 2 Burkitt lymphoma (Tables 1 and S1). Curative treatment intent was applied for 251 patients (90%), “watch and wait” and supportive care strategies were decided for 19 (7%) and 8 patients (3%), respectively. Among all patients, 28 (10%) were included in clinical trials, and 151 (54%) were included in the REALYSA program. The median rate of inclusion in REALYSA per month was 55% (range, 37% to 74%). Among nonincluded patients in REALYSA cohort, 39 (31%) refused to participate, 20 patients (16%) had emergency treatments, the clinician judged inclusion not feasible for 12 patients (9%), and the reason for noninclusion was unknown for 56 patients (44%) (Table S2).
TABLE 1.
Baseline clinical characteristics of patients according to REALYSA inclusion
| Clinical Characteristics | All Patients n = 278 (%) |
Included in REALYSA n = 151 (%) |
Nonincluded in REALYSA n = 127 (%) |
|---|---|---|---|
| Age ≥60 years | 168 (60%) | 82 (54%) | 86 (68%) |
| Gender: Male | 159 (57%) | 91 (60%) | 68 (54%) |
| Lymphoma Subtype | |||
| DLBCL | 104 (37%) | 59 (39%) | 45 (35%) |
| FL | 56 (20%) | 33 (22%) | 23 (18%) |
| HL | 46 (17%) | 26 (17%) | 20 (16%) |
| MZL | 35 (13%) | 15 (10%) | 20 (16%) |
| T-Cell Lymphoma | 19 (7%) | 6 (4%) | 13 (10%) |
| MCL | 16 (6%) | 10 (7%) | 6 (5%) |
| Burkitt lymphoma | 2 (1%) | 2 (1%) | 0 (0%) |
| Performance Status (n = 277) | |||
| 0-1 | 222 (80%) | 130 (86%) | 92 (72%) |
| 2-4 | 55 (20%) | 21 (14%) | 34 (27%) |
| Ann Arbor stage (n = 277) | |||
| I-II | 70 (25%) | 36 (24%) | 34 (27%) |
| III-IV | 207 (75%) | 115 (76%) | 92 (72%) |
| B symptoms (n = 277) | 90 (32%) | 42 (28%) | 48 (38%) |
| LDH level (UI/L) > Normal (n = 275) | 190 (68%) | 99 (66%) | 91 (72%) |
| Albumin level < 40 g/L (n = 267) | 145 (52%) | 66 (44%) | 79 (62%) |
| Charlson Comorbidity Index (CCI) | |||
| 0 | 178 (64%) | 100 (66%) | 78 (61%) |
| 1 | 49 (18%) | 28 (19%) | 21 (17%) |
| 2 | 25 (9%) | 12 (8%) | 13 (10%) |
| >2 | 26 (9%) | 11 (7%) | 15 (12%) |
| Previous solid cancer | 36 (13%) | 14 (9%) | 22 (17%) |
| ≥3 concomitant medications | 92 (33%) | 42 (28%) | 50 (39%) |
| Delay since the first symptoms (n = 275) | |||
| <3 months | 144 (52%) | 71 (47%) | 73 (58%) |
| 3-6 months | 73 (26%) | 50 (33%) | 23 (18%) |
| >6 months | 58 (21%) | 29 (20%) | 29 (23%) |
| Day of the first consultation or hospitalization | |||
| Monday-Thursday | 239 (86%) | 132 (87%) | 107 (84%) |
| Friday-Sunday | 39 (14%) | 19 (13%) | 20 (16%) |
| Access to Haematology Department (n = 276) | |||
| Hospitalization | 54 (19%) | 21 (14%) | 33 (26%) |
| Consultation | 222 (80%) | 130 (86%) | 92 (72%) |
| Distance from Hospital (km) (n = 277) | |||
| ≤15 | 117 (42%) | 57 (38%) | 60 (47%) |
| 16-59 | 95 (34%) | 61 (40%) | 34 (27%) |
| ≥60 | 65 (23%) | 33 (22%) | 32 (25%) |
| Therapeutic intention | |||
| Curative intent | 251 (90%) | 146 (97%) | 105 (83%) |
| Watch and Wait | 19 (7%) | 5 (3%) | 14 (11%) |
| Palliative intent | 8 (3%) | 0 (0%) | 8 (6%) |
Abbreviations: DLBCL, diffuse large B-cell lymphoma; FL, follicular lymphoma; HL, Hodgkin lymphoma; MZL, marginal zone lymphoma; MCL, mantle cell lymphoma; LDH, lactate dehydrogenase.
In multivariable analysis, curative treatment intent (OR=5.01; 95% confidence interval (CI), 1.55-16.23; P=0.006) and the presence of Ann Arbor III-IV stage (OR=2.08; 95% CI, 1-4.35; P=0.04) were positively associated with REALYSA enrolment. A trend for a decreased probability of REALYSA inclusion was observed for T-cell lymphoma and MZL subtypes (versus DLBCL subtype as reference), for patients with previous solid malignancies and for patients requiring immediate hospitalization (versus patients seen in consultation as reference) (Figure 1A and Table S3). For continuous variables, Figure 1B shows the relationship between age, albumin level and REALYSA inclusion probability. Patients from semirural areas (distance from residence to hospital 15-60 km) were more likely to participate than urban patients living near the hospital (<15 km) or patients in distant areas (Figure 1B). The probability of inclusion was similar regardless of the day of the week of the first patient visit, sex, comorbidities and performance status.
FIGURE 1.
Effect of covariates on the inclusion probability in REALYSA for the whole cohort summarized as odds ratios for categorical variables (A) and on the probability scale for continuous variables (B) as estimated with a generalized additive model.
Abbreviations: FL, follicular lymphoma; DLBCL, diffuse large B-cell lymphoma; HL, Hodgkin lymphoma; MZL, marginal zone lymphoma; MCL, mantle cell lymphoma; PS, performance status; ECOG, Eastern Cooperative Oncology Group; Friday-Sat, Friday-Saturday; Mon-Thu, Monday-Thursday; W&W, Watch and Wait; Supp. Care, Supportive Care; CRS, Cubic Regression Spline; LDH, lactate dehydrogenase.
Notes: Examples of results for Figure 1A: the odds of being included in REALYSA (versus not being included) increase by a factor of 2.08 for patients diagnosed with stage III-IV lymphoma, compared to patients diagnosed with stage I-II lymphoma, adjusted on all the other variables in the model; the odds of being included in REALYSA (versus not being included) increase by a factor of 0.82 for patients with an ECOG of 2-4, compared to patients with an ECOG of 1-2, adjusted on all the other variables in the model. Burkitt lymphomas were excluded from multivariable analyses because of an insufficient number of cases (n=2). The odds ratios with 95% confidence intervals are presented in Table S3.
In subgroup analyses for DLBCL treated with curative intent (n = 100), a trend was shown between a Charlson score >2 (OR=0.13, 95% CI, 0.02-1.13; P=0.06), the need for debulking chemotherapy (OR=0.20, 95% CI, 0.03-1.49; P=0.11) and a decreased inclusion probability (Figure S1 and Table S4). For FL patients, compressive syndrome was negatively associated with REALYSA inclusion (OR=0.10, 95% CI, 0.01-0.78; P=0.02) (Figure S1 and Table S5). For HL patients, the nodular sclerosis subtype was associated with REALYSA inclusion compared to other histological subtypes (OR=0.14, 95% CI, 0.02-0.91; P=0.03) (Figure S1 and Table S6). We did not find a significant difference in EFS according to REALYSA participation for DLBCL, FL and HL patients treated with curative intent (Figure S2).
We showed that the REALYSA program presented an enrolment rate of 54% in a tertiary haematology department, which is higher than the inclusion rate in clinical trials. We observed that patients with a clinically aggressive disease, especially those who needed immediate hospitalization, debulking chemotherapy (DLBCL subtype), palliative intent, low albumin level, bulky disease (FL subtype), and T-cell lymphoma, tended to be less included, a trend already described in clinical trials.8 Correlation with inclusion rate and age or comorbidities (DLBCL subtype) are also observed in clinical trials.9 We observed another profile of patients with a lower inclusion rate in the REALYSA: these patients had early stage, indolent histology (for instance MZL) and a watch and wait strategy. We hypothesized that physicians did not systematically propose REALYSA or that these patients with asymptomatic disease and managed in consultation did not feel concerned by this program. We showed that the correlation between the inclusion rate and the distance between residence and hospital had a specific distribution with a higher probability of inclusion of semirural patients. We hypothesize that these patients may feel more concerned about environmental exposures, which are research questions in REALYSA.6 These data need to be confirmed, and it will be interesting to integrate some socioeconomic variables known to impact the inclusion rate in clinical trials.10
Limitations of our study are the low number of studied patients, especially for subgroup analyses, and the short period of analysis. Our study should be viewed as a first step to better understand selection bias in the RW lymphoma cohort. Given that the REALYSA population represented only 54% of new lymphoma patients, caution should be taken regarding the generalizability and representativeness of the results. Our study could be extended in a broader lymphoma population using data from additional haematology department in order to better understand participation bias in real-life cohort. A special effort to improve knowledge about patient and clinician motivations for RW cohort participation would be also needed.
Supplementary Material
FUNDING INFORMATION
The REALYSA study is funded by several commercial organizations (Roche, Takeda, Janssen, Amgen, Celgene-BMS).
Footnotes
Presented in Poster form as part of the 64th Annual Meeting of the American Society of Hematology (ASH), New Orleans, Louisiana, December 9-12, 2022.
CONFLICT OF INTEREST STATEMENT
The authors have no conflicts of interest to declare.
ETHICS STATEMENT
This study was conducted in accordance with the Helsinki Declaration and approved by health authorities (French Data Protection Agency #2208143 and HDH publication #F20220705084800).
DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.
REFERENCES
- 1.Murthy VH, Krumholz HM, Gross CP. Participation in cancer clinical trials: race-, sex-, and age-based disparities. JAMA. 2004;291(22):2720–6. [DOI] [PubMed] [Google Scholar]
- 2.Booth CM, Tannock IF. Randomised controlled trials and population-based observational research: partners in the evolution of medical evidence. Br J Cancer. 2014;110(3):551–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.El-Galaly TC, Cheah CY, Villa D. Real world data as a key element in precision medicine for lymphoid malignancies: potentials and pitfalls. Br J Haematol. 2019;186(3):409–19. [DOI] [PubMed] [Google Scholar]
- 4.Elting LS, Cooksley C, Bekele BN, et al. Generalizability of cancer clinical trial results: prognostic differences between participants and nonparticipants. Cancer. 2006;106(11):2452–8. [DOI] [PubMed] [Google Scholar]
- 5.Sherman RE, Anderson SA, Dal Pan GJ, et al. Real-World Evidence - What Is It and What Can It Tell Us? N Engl J Med. 2016;375(23):2293–7. [DOI] [PubMed] [Google Scholar]
- 6.Ghesquières H, Rossi C, Cherblanc F, et al. A French multicentric prospective prognostic cohort with epidemiological, clinical, biological and treatment information to improve knowledge on lymphoma patients: study protocol of the « REal world dAta in LYmphoma and survival in adults » (REALYSA) cohort. BMC Public Health. 2021;21(1):432. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Wood SN. Generalized Additive Models: An Introduction with R, Second Edition. 2e éd. New York: Chapman and Hall/CRC. 2017; p.496. [Google Scholar]
- 8.Maurer MJ, Ghesquières H, Link BK, et al. Diagnosis-to-Treatment Interval Is an Important Clinical Factor in Newly Diagnosed Diffuse Large B-Cell Lymphoma and Has Implication for Bias in Clinical Trials. J Clin Oncol. 2018;36(16):1603–10. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Loh Z, Salvaris R, Chong G, et al. Evolution of eligibility criteria for diffuse large B-cell lymphoma randomised controlled trials over 30 years. Br J Haematol. 2021;193(4):741–9. [DOI] [PubMed] [Google Scholar]
- 10.Unger JM, Hershman DL, Albain KS, et al. Patient income level and cancer clinical trial participation. J Clin Oncol. 2013;31(5):536–42. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The data that support the findings of this study are available from the corresponding author upon reasonable request.