Level of evidence for FDA drug approvals in pivotal clinical trials of hematological malignancies

Clinical trials are integral to improve treatment outcomes for patients with hematological malignancies. The approval of drugs to treat hematological malignancies follows a process established by the United States Food and Drug Administration (FDA) to ensure a balance between efficacy and toxicity. The FDA approved more than 50 new drugs to treat various hematological malignancies in the last 5 years, with many of these drugs having been approved for the first time in the USA [1,2].

Early phase (I/II) clinical trials may provide evidence of clinical efficacy; however, the main goal of early phase trials is to assess safety. Results of phase III clinical trials provide the strongest evidence to support the use of new cancer medications. There are multiple concerns that result from the use of early phase clinical trials to support drug approvals [3]. Some examples include the lack of a control group [4], patient heterogeneity [5], short follow-up time due to bias related to early publication [6], and limitation to the generalizability of treatment results [2,7]. Moreover, the chosen surrogate endpoint may not actually predict clinical benefit. It is estimated that two-thirds of the FDA contemporary cancer drug approvals were based on surrogate endpoints [8]. However, up to 86% of approvals based on surrogate endpoints subsequently fail to show gains in overall survival (OS) or have unknown effects on survival [8].

We conducted a retrospective analysis of level of evidence supporting FDA drug approvals for hematological malignancies. Data on product labeling are available publicly at Drugs@FDA. Drugs approved from January 2016 until May 2020 were analyzed. The studied hematologic malignancies include multiple myeloma, leukemias, non-Hodgkin lymphomas, Hodgkin lymphoma, and myelodysplastic syndrome. We calculated the frequency of approvals based on early evidence clinical trials across different types of hematological malignancies.

Publicly accessible FDA reviews of 52 clinical trials supporting 49 drug approvals in the 5-year period were available for level of evidence review. Phase III/IV clinical trials supported 60% while earlier phase trials supported 40% of subsequent FDA hematological malignancies approvals. The level of evidence to support FDA approvals improved with time, with 50% of approvals in 2016 and 2017 supported by phase III clinical trials compared to 69% in 2019 (Figure 1). The median number of patients of phase III/IV clinical trials in the studied period was 326 with no significant change with time. Progression-free survival (PFS) was the primary endpoint in 58% and OS was the primary endpoint in 16% of all phase III/IV clinical trials in the studied period. The phase III/IV trials were routinely designed to detect a specified improvement with a false positive rate of at most 5% (p = 0.05) and a false negative rate of 10%–20% (power = 80%–90%).

Figure 1.

Trend of high-level evidence to support FDA approval of hematological malignancies. Percentage of high-level evidence approval with 95% confidence intervals.

Approvals in the studied period were based on early phase trials in mantle cell lymphoma (100%), diffuse large B-cell lymphoma (100%), chronic myeloid leukemia (75%), classic Hodgkin lymphoma (67%), and acute myeloid leukemia (56%). Eighteen drug approvals (37% of all approvals) were based on 13% of the total number of patients treated in the studied period. Most of the phase II clinical trials were single-arm (i.e. non-randomized). Only 3 of 17 (18%) phase II clinical trials were randomized (Table 1).

Table 1.

Frequencies of analyzed clinical trials.

Disease	Number of trials	Number of patients	Range of patient numbers enrolled in all studies	Median (range) of number of patients enrolled on high-level evidence trials	Number (%) of early phase studies that supported approval
Leukemia	26	7262	34–717	389 (80–717)	11 (42)
Chronic lymphoid leukemia	8	3094	106–535	432 (310–535)	1 (13)
Chronic myeloid leukemia	4	854	51–487	487 (n/a)	3 (75)
Acute lymphoid leukemia	4	922	75–405	366 (326–405)	2 (50)
Acute myeloid leukemia	9	2312	34–717	290 (247–717)	5 (56)
Hairy cell leukemia	1	80	n/a	80 (n/a)	0 (0)
Lymphoma	18	5212	80–1334	321 (83–1334)	9 (50)
Hodgkin lymphoma	3	1639	95–1334	1334 (n/a)	2 (67)
Diffuse large B-cell lymphoma	3	281	80–108	n/a	3 (100)
Mantle cell lymphoma	2	210	86–124	n/a	2 (100)
Follicular lymphoma	6	2300	83–1202	321 (83–1202)	1 (17)
Primary mediastinal B-cell lymphoma	1	53	n/a	n/a	1 (100)
T-cell lymphoma	3	729	131–372	226 (131–372)	0 (0)
Multiple myeloma	7	3543	122–1085	515 (263–1085)	1 (14)
Myelodysplastic syndrome	1	229	n/a	229 (n/a)	0 (0)
Total	52	16,246	34–1334	34–1334	21 (40)

n/a: not applicable.

Our study has limitations. First, we included only clinical trials that led to FDA approval. In addition, some of the hematological malignancies included are rare, which can limit the ability to perform large phase III clinical trials. Nonetheless, even for more frequent diseases, such as acute myeloid leukemia, with 19,940 new cases expected to occur in 2020 [9], only 44% of new drug approval were based on high-level evidence clinical trials in the last 5 years. The remaining, early phase-based drug approvals were based on small patient sample sizes.

In conclusion, evidence to support drug approvals in hematological malignancies was based on early phase trials data in more than a third of the times. Although early phase studies are appropriate to assess safety signals, further clinical activity assessment should be done to support the use of new drugs to treat hematological malignancies, as previous successful early phase studies have at times failed to show clinical activity in subsequent phase III studies. Although the use of newly approved drugs based on early phase studies evidence may be necessary (especially when the targeted diseases are rare), patients and healthcare providers should be aware of these caveats when using newly approved medications.