Demographic Differences Between Unselected Patients and Participants of Multiple Myeloma Clinical Trials in the US: A Threat to External Validity

Background: The use of clinical trial results to guide clinical practice is based on the assumption that results are externally valid and clinical trial subjects are a representative sample of patients to whom results can be applied. Multiple myeloma (MM) is primarily a disease of the elderly, and both tolerance to therapy and overall prognosis are influenced by age. MM biological characteristics and possibly response to therapy are different between Whites and racial-ethnic minorities (REM), particularly Blacks. We analyzed recent MM trials performed in the US to assess to what extent they included a representative sample of the MM population.

Methods: We retrieved all therapeutic clinical trials performed exclusively in the US and published between 2007 and 2014 for information on age, gender and racial-ethnic distribution of subjects. For the latter, subjects were classified as Whites or REM (defined as Hispanic and/or non-White). Data on unselected patients with newly diagnosed MM from the Surveillance Epidemiology and End Results (SEER-18) program were used to estimate the expected proportion of REM and each gender for the clinical trial population.

Results: We identified 128 published MM trials (8,869 subjects). Age of subjects was reported in 127 (99.2%) and gender distribution in 120 (93.8%) of trials, but racial-ethnic composition was reported in only 51 (39.8%) of trials (4,853 subjects). Trials reporting accrual of REM were larger (median 54 subjects, IQR 34.5-84.5) than other trials (median 35, IQR 24-50), P=0.004. Among industry-sponsored trials a higher proportion (57.1 %) reported REM accrual, than among NCI-sponsored (41.2%) or investigator-sponsored (29.0%) trials (overall P=0.013, P=0.003 for investigator vs. industry comparison). First line therapy trials had similar likelihood of reporting on accrual of REM (33.3%) than trials enrolling relapsed/refractory subjects (44.9%) or a mixed population (34.6%), P=0.4. Among subjects in the 51 trials that reported accrual of REM, 19.1% were minorities vs. expected 36.7% (ratio 0.52, 95% C.I. 0.49-0.55). In investigator-sponsored trials, the proportion of minorities was 15.9% (ratio 0.43, 95%CI 0.37-0.49) while among NCI-sponsored trials the proportion was 18.7% (ratio 0.50, 95% C.I. 0.45-0.50) and among industry-sponsored trials the proportion was 22.0% (ratio 0.61, 95% C.I. 0.55-0.68). Among subjects in the 120 trials that reported gender distribution, 56.7% were man vs. expected 56.9% (ratio 1.02, 95% C.I. 0.99-1.05).Median of median age of subjects enrolled in clinical trials was 61 years, in sharp contrast with the median age for MM diagnosis during the period that was 69 years. Only 9 (7.1%) trials the median age of subjects enrolled was higher than 65 years.

Conclusions: The participation of REM in MM trials is greatly under reported, particularly among investigator-sponsored trials. REM and older individuals are underrepresented in MM trials, potentially compromising external validity of resuts.l