Evaluation of Toxicities Related to Novel Therapy in Clinical Trials for Women With Gynecologic Cancer

Abstract

BACKGROUND:

Women with gynecologic cancer may be at increased risk for adverse events (AEs) due to peritoneal disease burden and prior treatment (surgery, chemotherapy, and pelvic radiotherapy). This study compared the toxicity profiles of patients with and without gynecologic cancer enrolled in phase 1 trials.

METHODS:

This was a retrospective analysis of the National Cancer Institute phase 1 database for all trials enrolling 1 or more patients with gynecologic cancer over 2 decades (1995-2015). Clinical parameters collected included demographics, cancer history, trial information, AEs, and responses. AEs (according to the Common Terminology Criteria for Adverse Events) were documented for each patient during treatment, and they were counted once and analyzed on the basis of the highest grade and drug attribution. Multiple regression models were used to compare AEs at the baseline and during treatment.

RESULTS:

A total of 4269 patients enrolled in 150 trials were divided into 3 groups: 1) women with gynecologic cancer (n = 685), 2) women with nongynecologic cancer (n = 1698), and 3) men with cancer (n = 1886). The median age was 58 years. The mean number of total AEs reported during treatment was highest for women with gynecologic cancer (17.1 vs 14.7 vs 13.5; P < .001), even though they were similar at the baseline (7.0 vs 7.4 vs 7.0; P = .09). The mean number of drug-related AEs was also highest for women with gynecologic cancer (8.3 vs 6.9 vs 6.2; P < .001). Grade 3 to 5 AEs were similar (2.3 vs 2.3 vs 2.1); however, grade 2 AEs were more frequent in women with gynecologic cancer (4.6 vs 3.9 vs 3.5). Treatment discontinuations due to AEs were similar (9% vs 9% vs 10%).

CONCLUSIONS:

Women with gynecologic cancer experienced more frequent low-grade AEs during treatment, and this warrants attention to support their symptom burden. Study dose management should be considered for recurrent grade 2 AEs, particularly during continuous therapy.

INTRODUCTION

The systematic evaluation of therapeutic agents through clinical trials has enabled the development of effective cancer treatments that improve patient outcomes. Following The Cancer Genome Atlas molecular characterization of solid tumors, including gynecologic cancers, many different targeted therapies have been developed, and some, such as bevacizumab and PARP inhibitors, are now approved for use in the routine treatment of gynecologic cancers.¹ Stringent protocols are in place to report all adverse events (AEs) that occur in patients participating in all phases of clinical trials so that we can better understand the toxicity profiles of these investigational agents. Phase 1 clinical trials are primarily intended to assess the safety and tolerability of investigational agents.^2-4 The value of phase 1 trials in evaluating AEs and accurately defining drug-related toxicity profiles has been well documented.⁵

Patients with advanced gynecologic cancers who are enrolled in clinical trials may be at increased risk for certain toxicities because of the extent of peritoneal disease and prior combined treatments, including extensive cytoreductive surgery, chemotherapy, and radiotherapy. An example would be the phase 2 study investigating BKM120, a PI3K inhibitor, for endometrial cancer, which was prematurely discontinued because of an unfavorable toxicity profile observed in this specific population.⁶ Although BKM120 was established as tolerable with an acceptable safety profile at the estimated maximal tolerated dose of 100 mg/d,^7,8 the same dose resulted in toxicities for women with recurrent endometrial cancer that were not reported in the initial phase 1 trial. Depending on the primary tumor sites and even histological subtypes, the safety profile of BKM120 was found to vary significantly.⁶ Therefore, our study investigated AEs documented in phase 1 studies of patients across all tumor types, and we compared the toxicity profiles of patients with gynecologic cancers with those of patients with other tumor types. In addition, we explored whether there were sex differences in toxicity profiles.⁹ We aimed to identify the prevalent AEs experienced by women with gynecologic cancers and thereby inform the design of prospective studies and supportive care for these women in clinical practice.

MATERIALS AND METHODS

A retrospective review was conducted of all patients who participated in phase 1 studies from 1995 to 2015 via the National Cancer Institute (NCI) database. The NCI database includes each and every phase 1 trial conducted by NCI investigators since the inception of the database in the year 1995. The lead principal investigator and local investigators for each phase 1 trial registered in the database were responsible for enrolling patients who met the eligibility criteria. In this review, only trials that enrolled at least 1 patient with gynecologic cancer were selected to compare their toxicity profile with those of other cancer types. These patients had received at least 1 dose of the clinical trial drug, and their data entry was complete from the baseline to the end of trial participation. The unlinked data set was extracted by the NCI with unique patient number labeling. The study eligibility was determined and the data were analyzed by Y.C.L. and L.W. Senior authors E.C.K. and S.L. oversaw all study procedures and verified that the data presented were accurate. All selected patients were divided into 3 groups: 1) women with gynecologic cancers, 2) women with nongynecologic cancers, and 3) men with cancers. Women with gynecologic cancers were defined as women who had ovarian/fallopian tube/peritoneal cancers, uterine cancers, cervical cancers, or vulvar cancers. All patients had been consented for data collection for their respective clinical trial and registered in the NCI database using an unlinked dataset. NCI consent templates have addressed future unspecified use of unlinked data. Accordingly, this research pertained to an NCI exempt research category.

Information on phase 1 trials, patient demographics, cancer histories, treatments, and AEs was extracted for analysis. Clinical data included the age at study enrollment, sex, primary cancer diagnosis, histological subtype, prior lines of treatment, treatment cycles, AEs, best treatment response, duration on trial, and reason for being off study. AEs at the baseline and during treatment for each patient were coded according to the trial’s protocol-designated versions of the Common Toxicity Criteria or Common Terminology Criteria for Adverse Events (CTCAE). All uncoded AEs (predominantly from earlier trials) were manually reviewed and categorized accordingly. Recurring AEs for each patient were counted as 1 AE for the whole duration of treatment and were analyzed on the basis of the highest grade and drug attribution.

Analyses were performed with SAS 9.4 (SAS Institute, Cary, North Carolina). Multivariable linear regression models were used to compare the averages of total AEs at the baseline and during treatment among the 3 groups, and they controlled for the number of treatment cycles and the year of treatment. All P values presented are 2-sided.

RESULTS

Patient Characteristics

A total of 356 phase 1 trials were identified in the NCI phase 1 database between 1995 and 2015; 159 of these trials included patients with gynecologic cancer in their eligibility criteria (Fig. 1). After data extraction, there were 150 trials with at least 1 patient with gynecologic cancer enrolled. A total of 4317 patients with baseline information were enrolled in these trials. There were 48 patients excluded because of incomplete treatment information. The remaining 4269 patients were divided into 3 groups: 1) women with gynecologic cancers (n = 685), 2) women with nongynecologic cancers (n = 1698), and 3) men with cancers (n = 1886).

Figure 1.

Study trial and patient selection flow diagram. Gyne indicates gynecologic; NCI, National Cancer Institute; nongyne, nongynecologic.

The patient characteristics are displayed in Table 1. The median age of all patients at the time of study enrollment was 58 years (range, 18-87 years), and the top 6 primary cancers were gastrointestinal (n = 1153 [27%]), gynecologic (n = 685 [16%]), breast (n = 475 [11%]), hepatobiliary/pancreatic (n = 429 [10%]), lung (n = 349 [8%]), and genitourinary (n = 330 [8%]). The majority of gynecologic cancers were ovarian/fallopian tube/peritoneal cancers (n = 527 [77%]), which were followed by cervical cancer (n = 76 [11%]) and uterine cancer (n = 72 [11%]). The median number of lines of treatment before study enrollment for all patients was 4 (range, 1-20).

TABLE 1.

Patient Characteristics at Enrollment in Phase 1 Trials

Characteristic	Women With Gynecologic Cancer (n = 685)	Women With Nongynecologic Cancer (n = 1698)	Men With Cancer (n = 1886)
Age, median (range), y	56 (21-84)	56 (18-86)	60 (20-87)
Prior lines of therapy, No. (%)
1	67 (10)	181 (11)	245 (13)
2-3	185 (27)	504 (30)	704 (37)
≥4	421 (61)	916 (54)	823 (44)
Investigated agents, No. (%)
Targeted therapy	528 (77)	1259 (74)	1271 (67)
Cytotoxic	116 (17)	375 (22)	541 (29)
Immunotherapy	10 (1)	11 (1)	14 (1)
Other	31 (5)	53 (3)	60 (3)

Trial Characteristics

Nearly all phase 1 studies (n = 145 [97%]) included patients with advanced cancers across multiple tumor sites. The remaining 5 trials recruited patients with gynecologic cancers only for a total of 49 study participants. As shown in Table 2, therapeutic agents were classified as targeted therapy (n = 97 [65%]), cytotoxic agents (n = 35 [23%]), or immunotherapy (n = 13 [9%]). Targeted therapy included DNA damage repair pathway modifiers (n = 29 [19%]) such as PARP inhibitors and heat shock protein inhibitors, epigenetic modifiers (n = 13 [9%]) such as histone deacetylase inhibitors and DNA methyltransferase inhibitors, and cell cycle checkpoint regulators (n = 10 [7%]) such as cyclin-dependent kinase inhibitors and checkpoint kinase 1 inhibitors. The main cytotoxic agents investigated were microtubule inhibitors (n = 13 [9%]) and topoisomerase inhibitors (n = 10 [7%]), and the predominant immunotherapy agents were vaccine therapies (n = 8 [5%]).

TABLE 2.

Phase 1 Trial Characteristics and Classes of the Investigated Agents

Characteristic	Trials (n = 150), No. (%)
Trials by year
1995-1999	49 (33)
2000-2004	38 (25)
2005-2009	29 (19)
2010-2015	35 (23)
Investigated agents
Targeted therapy
DNA damage repair pathway	29 (19)
Epigenetic modification	13 (9)
Cell cycle checkpoint regulators	10 (7)
VEGF/VEGFR inhibition	9 (6)
BRAF-MEK-ERK pathway	7 (5)
NF-κB signaling pathway	6 (4)
Notch signaling pathway	6 (4)
PI3K/AKT/mTOR pathway	4 (3)
Apoptosis pathway	4 (3)
Hormone receptor pathway	3 (2)
Protein kinase C pathway	3 (2)
EGFR inhibition	2 (1)
Hedgehog signaling pathway	1 (1)
Cytotoxic
Microtubules	13 (9)
Topoisomerase inhibitors	10 (7)
Alkylating agents	6 (4)
Antimetabolites	2 (1)
HIF-1α	2 (1)
Platinum	1 (1)
Retinoid	1 (1)
Immunotherapy
Vaccine therapy	8 (5)
Interleukin	3 (2)
PD-1	1 (1)
IDO1 inhibitor	1 (1)
Other
Miscellaneous agents	5 (3)

Reports of AEs

The average numbers of total AEs reported at the baseline were similar across all 3 groups (7.0 for women with gynecologic cancer vs 7.4 for women with nongynecologic cancer vs 7.0 for men with cancer; P = .09). However, the average numbers of total AEs reported during treatment were significantly different in the 3 groups (17.1 vs 14.7 vs 13.5; P < .001) after we accounted for the number of treatment cycles and the year of treatment (Fig. 2). During treatment, the average number of total AEs attributable to the drug was also higher (8.3 vs 6.9 vs 6.2; P < .001; Table 3). Table 3 shows AE severity and drug attribution. The average numbers of all grade 3 to 5 AEs were similar across the 3 groups (2.3 vs 2.3 vs 2.1). There was a higher average number of grade 1 AEs reported in women with gynecologic cancer (10.13 vs 8.41 vs 7.89; P < .001). There was also a higher average number of grade 2 AEs reported in women with gynecologic cancer (4.6 vs 3.9 vs 3.5; P < .001). More specifically, the number of grade 2 AEs attributable to the drug on treatment was also higher (2.36 vs 1.82 vs 1.61). There was a numerically higher average number of total AEs for abdominal-related AEs and myelosuppression (Table 3).

Figure 2.

Number of AEs reported (A) at the baseline and (B) during treatment during participation in phase 1 trials. AE indicates adverse event; CI, confidence interval; gyne, gynecologic; nongyne, nongynecologic; Rx, treatment.

TABLE 3.

Adverse AEs per Patient Reported During Treatment in Phase 1 Trials

Characteristic	Women With Gynecologic Cancer (n = 685)	Women With Nongynecologic Cancer (n = 1698)	Men With Cancer (n = 1886)
No. of AEs
Average AEs at baseline	7.0	7.4	7.0
Average AEs on treatment	17.1	14.7	13.5
AEs (on treatment only)
Attributable to drug	8.31	6.86	6.20
Not attributable to drug	8.76	7.79	7.32
Grade of AEs (on treatment only)
1	10.13	8.41	7.89
2	4.64	3.95	3.51
3-5	2.30	2.30	2.10
Grade of AEs (on treatment only and attributable to drug)
1	4.76	3.94	3.65
2	2.36	1.82	1.61
3-5	1.19	1.10	0.94
Categories of AEs (on treatment only)
Abdominal-related AEsa	5.54	4.49	4.26
Myelosuppression	2.06	1.76	1.50
Electrolyte derangement (Na, K, Mg)	0.86	0.71	0.62
Liver enzyme derangement	0.83	0.95	1.00
Genitourinary AEs	0.54	0.32	0.39
Infection	0.52	0.38	0.34

Abbreviation: AE, adverse event.

^aThese AEs include nausea, vomiting, constipation, diarrhea, abdominal pain, and so forth.

The 5 most prevalent AEs in women with gynecologic cancer were as follows: nausea (n = 617 [90%]), fatigue (n = 587 [86%]), anemia (n = 381 [56%]), anorexia (n = 357 [52%]), and vomiting (n = 355 [52%]). For women with nongynecologic cancer, similar trends but lower rates of these symptoms were noted apart from anemia (n = 451 [27%]): nausea (n = 1373 [81%]), fatigue (n = 1372 [81%]), anorexia (n = 780 [46%]), and vomiting (n = 719 [42%]). Men with cancer reported similar rates of fatigue (n = 1625 [86%]) and anorexia (n = 902 [48%]) but much lower rates of nausea (n = 1385 [73%]), vomiting (n = 661 [35%]), and anemia (n = 624 [33%]).

Treatment Responses

Women with gynecologic cancer remained on treatment marginally longer (cycles, 4.3 for women with gynecologic cancer vs 3.3 for women with nongynecologic cancer vs 3.1 for men with cancer; P < .001) and achieved higher objective response rates (11% vs 6% vs 3%; P < .001; Table 4). The majority of patients discontinued treatment because of disease progression (62% vs 68% vs 69%; Table 4). The treatment discontinuation rates due to AEs were similar (9% vs 9% vs 10%; Table 4). There was a higher percentage of death during treatment for men versus women (4% vs 2%; P = .007; Table 4).

TABLE 4.

Patient Characteristics During Treatment in Phase 1 Trials

Characteristic	Women With Gynecologic Cancer (n = 685)	Women With Nongynecologic Cancer (n = 1698)	Men With Cancer (n = 1886)
Treatment cycles received, mean	4.3	3.3	3.1
Reason off trial, No. (%)
Disease progression	423 (62)	1150 (68)	1301 (69)
Adverse events	61 (9)	160 (9)	185 (10)
Death during treatment	15 (2)	41 (2)	76 (4)
Best response, No. (%)
Complete response	12 (2)	14 (1)	4 (—)
Partial response	65 (9)	91 (5)	57 (3)
Stable disease	246 (36)	523 (30)	591 (31)
Progressive disease	257 (38)	782 (46)	947 (50)

DISCUSSION

This study reports an analysis of the toxicity profiles of women with gynecologic cancers participating in phase 1 studies across the United States and Canada. The NCI’s collaborative network of multiple cancer centers has enabled better patient access to clinical trials and integration of research resources to advance the science and systematic investigation of novel therapeutics. We showed that women with gynecologic cancer experienced a greater number of AEs during treatment than patients with other cancer types. Specifically, they reported higher average numbers of total drug-related AEs and abdominal-related AEs, the majority of which were low-grade toxicities (grades 1 and 2). The most common AEs found in women with gynecologic cancer were abdominal symptoms, such as nausea, vomiting, anorexia, constipation, diarrhea, abdominal pain, and distension. There were also notable sex difference in the AE profile, with women being more likely to report nausea and vomiting despite similar rates of anorexia. Therefore, there should be specific consideration for trial safety monitoring and symptom management for women with gynecologic cancer.

Early-phase trial designs have become highly selective in their eligibility criteria as more targeted therapies are being examined in biomarker-specific populations; these designs often also use novel statistical methods to improve the efficiency of dose escalation to its maximal therapeutic dose.^4,10,11 As the pace of drug development increases and becomes more complex, greater attention should be paid to monitoring the safety of patients. The assessment of AEs has been traditionally based on the physician’s assessment and the seriousness of the events. There is growing emphasis on the need to take into account repetitive grade 2 toxicity events observed during continuous dosing in the definition of dose-limiting toxicity (DLT).^12-15 Information throughout the treatment cycles, specifically about intolerable clinical grade 2 toxicities leading to a reduction in patients’ relative dose intensity, should be incorporated when the recommended phase 2 dose is determined for further studies.¹⁵ The low-grade AEs observed in this study, particularly those attributable to the investigated agent during treatment, seem to be more prevalent in women with gynecologic cancers and may not be recognized as problematic. Study dose modifications, such as changing continuous therapy to intermittent dosing, could be considered for study patients experiencing recurrent grade 2 AEs.

The integration of patient-reported outcomes into clinical trials may provide greater insight into regimen tolerability, patients’ symptom burden, and the implications of AEs beyond the CTCAE grading by clinicians.¹⁶ A recent report by Henon et al¹³ showed that patients’ most feared AEs, which they deemed intolerable even when they were low grade (grade 1 or 2), were gastrointestinal and neurological toxicities as well as personality changes; these AEs were different from physicians’ most feared toxicities and the definition of DLT. To this end, the NCI has developed a patient-reported outcomes version of CTCAE for use in clinical trials and is actively addressing the operational concerns to allow its full integration into clinical trials in collaboration with the US Food and Drug Administration and the Office for Human Research Protection.^17-19 In addition, fit-for-purpose patient-reported outcome measures, such as the Measure of Ovarian Symptoms and Treatment (MOST), can be used in clinical trials specifically for women with recurrent ovarian cancer.²⁰ The incorporation of patient-reported outcomes into trial assessments may uncover low-grade AEs that could be addressed early and allow a better assessment of a patient’s toleration of the investigated drug.

Our large data set demonstrated that patients with gynecologic cancers in comparison with other cancer types had a higher average number of low-grade (grade 1 and 2) AEs and abdominal-related AEs, and this highlights the need to support their symptom profile and improve their quality of life. The most common AEs found were nausea, vomiting, anorexia, constipation, diarrhea, and abdominal pain and distension. In addition, fatigue was one of the most prevalent AEs reported among all groups of patients participating in the trials. This symptom burden is consistent with the literature and is perhaps underestimated, as evidenced by Friedlander et al’s report showing that almost half of patients with ovarian cancer reported a constellation of gastrointestinal symptoms contrary to their physician’s account of their AEs.²¹ Adaptive development of digital health technologies, such as the use of a mobile application for regular monitoring of AEs and protocol-driven feedback interventions, may facilitate closer symptom management and early interventions to ameliorate the degree of toxicity experienced.²²

Effective novel agents are required to improve outcomes and to expand the existing limited therapeutic options for women with gynecologic cancers. Patients participating in phase 1 trials typically have progressed through multiple lines of treatment, with a median of 4 lines of treatment in our study. Their participation in clinical trials is often hindered by concerns about the high abdominal disease burden and their risk of developing malignant bowel obstructions or fistulas. The eligibility criteria in current trial protocols typically exclude patients with a recent history of bowel obstruction. Our study showed that the main difference in toxicity outcomes for women with gynecologic cancers was the incidence of repetitive moderate-grade AEs, which can be addressed by early management and supportive care. Specific supportive care strategies could be incorporated into clinical trial protocols to address these risks. These may include the following: 1) proactive screening and management of bowel function; 2) avoidance or reduction in the use of medications prone to altering bowel function, such as magnesium replacement, which causes diarrhea, or 5-hydroxytryptamine inhibitor anti-emetic agents, which cause constipation; 3) patient education to increase awareness as well as early recognition and reporting of abdominal symptoms to lead to early interventions; 4) screening for diet and nutrition; and 5) encouragement to exercise and take regulated rest periods to cope with cancer fatigue.

Our data set consists of large, comprehensive phase 1 trials that were rigorously conducted by the NCI at specialized cancer centers since 1995. The group analysis was intended to elucidate overall trends, and it would be difficult to account for individual nuances, such as changes in the nature of drug therapy from cytotoxic agents to targeted agents; it does not reflect patients on immunotherapy agents. The relatively short DLT evaluation period and generally short follow-up time of phase 1 trials limit the assessment of long-term toxicities. The collection of patient-reported outcomes in prospective clinical trials may provide greater insight into the full impact of low- and intermediate-grade toxicities over time and add value beyond the initial close monitoring of DLT.

In conclusion, our large study highlighted the AE profiles of women with gynecologic cancers while they were participating in phase 1 studies. These women experienced greater low-grade AEs during treatment in comparison with women or men with other cancer types, so specific supportive care recommendations are warranted. Our findings also reaffirm the notion that novel therapeutics should be studied specifically for gynecologic cancers in early-phase trials to allow an in-depth understanding of drug tolerability and toxicity profiles.