Adoptive cell therapy in gynecologic cancers: A systematic review and meta-analysis

Abstract

Adoptive cell therapy (ACT) has shown promise in hematologic and solid tumors. While data supports immunogenicity of gynecologic cancers, the benefit of ACT is not yet clear. To address this question, we performed a comprehensive systematic review and meta-analysis. Eligible studies included those reporting oncologic response or toxicity data in at least one patient with any gynecologic cancer treated with ACT. Chi-square test and multivariable logistic regression were performed to identify predictors of response. We retrieved 281 articles, and 28 studies met our inclusion criteria. These comprised of 401 patients including 238 patients with gynecologic cancers (61.8% ovarian, 34.0% cervical, 2.9% endometrial, and 1.2% other). In patients with gynecologic cancers, response rates to ACT were 8.1% complete response, 18.2% partial response, and 31.4% stable disease, for an objective response rate (ORR) of 26.3%, disease control rate (DCR) of 57.6%, and median response duration of 5.5 months. Patients in studies reporting ≤1 median line of prior therapy had a higher ORR (52.9% vs. 22.6% for >1, p<0.001), although DCR in the >1 group was still 53.2%. ORRs by ACT type were tumor infiltrating lymphocytes (TIL) 41.4%, natural killer cells 26.7%, peripheral autologous T-cells 18.4%, T-cell receptor-modified T-cells 15.4%, and chimeric antigen receptor T-cells 9.5% (p=0.001). ORR was significantly improved with inclusion of lymphodepletion (34.8% vs. 15.4% without, p=0.001). On multivariable analysis controlling for cancer type and lymphodepletion, TIL therapy was predictive of objective response (odds ratio 2.6, p=0.011). The rate of grade 3 or 4 toxicity was 46.0%. All grade adverse events included fever, hypotension, dyspnea, confusion, hematologic changes, nausea/vomiting, fatigue, and diarrhea. In conclusion, ACT is a promising treatment modality in gynecologic cancer. We observed a particular benefit of TIL therapy and suggest inclusion of lymphodepletion in future trials.

Introduction

Cancer care has been revolutionized by the advent of immunotherapy. Adoptive cell therapy (ACT) is a type of immunotherapy that has shown favorable results, particularly for hematologic malignancies. For recurrent or advanced gynecologic cancers, few standard treatment options exist, and their efficacy tends to be limited. While ACT is being increasingly incorporated into the treatment of solid tumors, its role in the management of gynecologic cancers remains to be explored.

ACT exploits physiologic immunosurveillance mechanisms against cancer. Cancer patients develop both innate and adaptive immune responses to their cancer, and an array of immunogenic tumor antigens has been identified.¹ This concept is bolstered by the correlation between the presence of tumor-infiltrating lymphocytes (TIL) and positive oncologic prognosis in various tumor types,^2–4 including ovarian cancer.^5–7 T-cell-mediated anticancer response starts with T-cell receptor recognition of the tumor antigen peptide in combination with the human leukocyte antigen (HLA). This recognition is followed by the interaction of the costimulatory surface molecules B7 and CD28, which leads to T-cell activation, trafficking to the tumor, and immune-mediated killing of cancer cells. In the “escape” phase, tumor variants with genetic and epigenetic changes that confer resistance to detection or elimination—such as the creation of an immunosuppressive milieu—aid in cancer cell expansion and lead to clinical detection.¹ In ACT, autologous immune cells are extracted, expanded, stimulated or engineered ex vivo, and reinfused for therapeutic purposes, often along with growth factors such as interleukin-2 (IL-2) to augment in vivo anticancer activity. ACT can use peripheral autologous T-cells, TIL, T-cell receptor-modified T-cells (TCR), chimeric antigen receptor T-cells (CAR T), and natural killer (NK) cells. Peripheral autologous T-cells and TIL therapy derive T-cells from the peripheral blood and tumor, respectively, which are then expanded and stimulated prior to reinfusion.⁸ Engineered TCR and CAR T therapy were developed in an effort to improve the therapeutic efficiency and use T-cells that are genetically modified to encode receptors recognizing cancer-specific antigens. CAR T and NK cell therapy are advantageous in that they are not HLA-restricted and may allow off-the-shelf allogenic therapy.^9,10

Use of ACT in solid tumors has shown remarkable success in melanoma, with objective response rates (ORRs) as high as 60%.¹¹ The addition of preconditioning lymphodepletion, which removes competing and/or inhibitory endogenous lymphocytes, improved the durability of response: the mean duration was greater than 11 months,^12,13 and one study reported an ongoing complete response rate of 22% at 3 years post-treatment.¹⁴ There is mounting evidence that many gynecologic cancers are immunogenic,¹⁵ and the use of ACT in this setting warrants further investigation. The purpose of this meta-analysis was to investigate the benefit of ACT in gynecologic cancers as it pertains to oncologic outcomes, molecular targets, and toxicity to reveal areas requiring further study and refinement.

Methods

We performed a comprehensive systematic search of Ovid MEDLINE, Ovid EMBASE, Cochrane Library, Web of Science, as well as ClinicalTrials.gov and International Clinical Trials Registry. In addition, we searched “gray literature” resources from conferences, dissertations, reports, and other relevant citations. Searches were restricted to human subjects and English language articles. Search structures, subject headings, and keywords were tailored to each database by a medical research librarian (KJK) specializing in systematic reviews. MeSH, Emtree, and keywords were searched to identify concepts related to gynecologic cancer and ACT. The full search strings for all databases can be found in the supplementary section (S1).

After the initial search, Rayyan was used to screen citations.¹⁶ Two of the principal authors (JS and MN), who were blinded to each other’s selections, independently screened the titles and abstracts to identify potentially relevant studies. After unblinding, disagreements were resolved by consensus discussion as outlined in the Cochrane Handbook,¹⁷ and a third arbitrator was not necessary. Studies that passed the title and abstract review were retrieved for full-text review using a pre-determined inclusion criteria. Eligible studies included those reporting standardized oncologic response and/or toxicity data in at least one patient with any gynecologic cancer treated with ACT. We excluded animal or in vitro studies, multiple reports of the same study, review articles, meta-analyses, editorials, and practice guidelines.

Study characteristics were collected, including types of ACT, phase of trial, treatment and lymphodepletion regimen, target antigen (if any), and criteria for tumor response and toxicity. Patient and response characteristics collected included types of cancer, number of patients by cancer type, number of prior treatments, number of patients experiencing treatment response, and duration of response. Toxicity variables included number of patients who reported grade 3 or 4 toxicity, dose limiting toxicity, fever, hypotension, dyspnea, and other commonly reported toxicities according to Common terminology criteria for adverse events (CTCAE).

Statistical analyses were performed using IBM SPSS version 26. For comparison of pooled patient-level data, Pearson-chi square test was used to examine differences in ORR, sum of complete response (CR) and partial response (PR), as well as disease control rate (DCR), sum of ORR and stable disease (SD). Missing data was assumed to be progressive disease (PD) for conservative estimation of ORR and DCR. Multivariable logistic regression was performed based on significant terms to identify predictors of response. All p values were two-sided with 0.05 as the level of statistical significance and 95% as the confidence interval.

Results

We retrieved 281 articles for review, of which 28 studies encompassing 32 study arms ultimately met inclusion criteria (Table 1). The PRISMA flow diagram shows the search and selection process (S2). In total, there was a total of 401 patients, and 238 of these patients had gynecologic cancers (61.8% ovarian, 34.0% cervical, 2.9% endometrial, and 0.8% vaginal, and 0.4% vulvar). All papers described the use of ACT in the treatment of recurrent or advanced cancer with a median range of 0–12 lines of prior therapy.

Table 1.

Eligible studies including at least one patient with gynecologic caner treated with adoptive cell therapy reporting standardized oncologic or toxicity data

Article	Clinical trial identifier	Study type	Type of ACT	Regimen	LD	Cancer types included	GYN (n)	GYN ORR (%)*
Steis 199046		I	PATC	IP LAK + IL2		Ovarian, endometrial, GI	11	18
Stewart 199047		I	PATC	IP LAK + IL2		Ovarian	10	0
Aoki 199141		I	TIL	Arm1: mono	√	Ovarian	7	71
				Arm2: +chemo incl cisplatin			10	90
Freedman 199448		Pilot	TIL	IP + IL2		Ovarian	8	0
Canevari 199538		II	PATC	IP CD3 folate receptor target + IL2		Ovarian	26	23
Kershaw 200625		I	CAR T	Arm1: Folate receptor target + IL2		Ovarian	8	0
				Arm2: Folate receptor and PBMC target			6	0
Geller 201118	NCT01105650	II	NK	Haploidentical NK + IL2	√	Ovarian, breast	14	29
Wright 201249		Pilot	PATC	IP CD8 MUC1 target		Ovarian	7	14
Kandalaft 201350,51	NCT00603460	I	PATC	+DC vaccine (freeze thaw and oxidized tumor), bevacizumab	√	Ovarian	8	13
Skeate 201352		CSR	NK	Haploidentical NK		Ovarian	1	0
Tanyi 201453	NCT01312376	I	PATC	+DC vaccine (oxidized whole tumor), bevacizumab	√	Ovarian	8	0
Han 201554		CSR	PATC	+DC vaccine (MASCT)		Cervical	1	(100)
Tanyi 201721		CSR	CAR T	Mesothelin target		Ovarian	1	-
Lu 201755	NCT02111850	I	TCR	CD4 MAGE A3 target + IL2	√	Cervical, various solid	3	33
Chen 201831	NCT02886897	I	PATC	CIK PD1 target		Ovarian, HNC, lung, breast, GI, GU	1	0
Pedersen 201830	NCT02482090	Pilot	TIL	+ IL2	√	Ovarian	6	0
Bryson 201956	NCT03578406	I	TCR	Arm1: HPV16 E6 target		Cervical	5	0
				Arm2: HPV16 E6 target + anti-PD1 scFV			2	0
Butler 201919	NCT02869217	I	TCR	TBI1301 NYESO-1 target	√	Ovarian, endometrial, sarcoma, melanoma	2	0
Doran 201957	NCT02280811	I/II	TCR	HPV16 E6 target + IL2	√	Cervical, vaginal, anal, HNC	7	0
Haas 201926	NCT02159716	I	CAR T	Mesothelin target	√a	Ovarian, mesothelioma, pancreatic	5	0
Hattori 201920	NCT02366546	I	TCR	NYESO-1 target	√	Ovarian, breast, HNC, melanoma, sarcoma	1	0
Jazaeri 201934 O’Malley 202133	NCT03108495	II	TIL	Arm1: LN145 + IL2	√	Cervical	27	44
				Arm2: LN145 + pembrolizumab, IL2		Cervical, HNC, melanoma	10	50
Qiao 201958	NCT03757858	I	PATC	Hyperthermia + pembrolizumab or chemo		Ovarian, cervical, endometrial, various solid	15	33
Stevanovic 201942	NCT01585428	II	TIL	HPV E6/E7 target + IL2	√	Cervical, vaginal, anal, HNC	19	26
Tsimberidou 201959	NCT02876510	I	PATC	IMA101 + IL2	√	Ovarian, various solid	1	0
Fang 202132	NCT03615313	CSR	CAR T	Mesothelin and PD1 target + apatinib		Ovarian	1	(100)
Hong 202122	NCT03907852	I	CAR T	TC210 mesothelin target	√	Ovarian, mesothelioma	1	(100)
Nagarsheth 202143	NCT02858310	I	TCR	HPV 16 E7 target	√	Cervical, vulvar, anal, HNC	6	50

^*study-level data for gynecologic patients only

^aTwo of 5 patients received lymphodepletion

ACT, adoptive cell therapy; CSR, case report; GI, gastrointestinal cancer; GU, urologic cancer; GYN, gynecologic cancer; HNC, head and neck cancer; IP, intraperitoneal delivery; LD, lymphodepletion; ORR, objective response rate; DCR, disease control rate. CAR T, chimeric antigen receptor T-cells; CIK, cytokine-induced killer cells; DC, dendritic cell; LAK, lymphocyte-activated killer T cells; NK, natural killer cells; PATC, peripheral autologous T-cells; TCR, T-cell receptor-modified T cells; TIL, tumor-infiltrating lymphocytes.

Table 2 shows study and patient characteristics. At the study level, the types of ACT used included peripheral autologous T-cell transfer (31.3%), TCR (21.9%), TIL (21.9%), CAR T (18.8%), and NK cells (6.3%). Of the 10 studies that included peripheral autologous T-cell transfer, 7 used ex vivo antigen, antibody, or dendritic cell stimulation. Most studies reported Phase I trials (57.1%) followed by Phase II trials (14.3%), as well as combined Phase I/II trials (3.6%), pilot studies (10.7%), and case reports (14.3%). With regards to preconditioning, 42.9% of the studies included IL-2 support, and 53.1% included lymphodepletion. The type of lymphodepletion regimen used included cyclophosphamide plus fludarabine (70.6%) and cyclophosphamide alone (29.4%). One study included low-dose whole-body radiation in addition to the cyclophosphamide plus fludarabine regimen.¹⁸ Of the patients who received IL-2, 65.0% also underwent lymphodepletion. Most of the studies used the Response Evaluation Criteria In Solid Tumors (RECIST) for reporting tumor response (85.2%); others used the Japanese Society for Cancer Therapy criteria, World Health Organization criteria, Eastern Cooperative Oncology Group criteria, and surgical measurements (3.7% each, data not shown). In general, study-level ORR seemed to improve over time in accordance with changes in regimen including the type of ACT and use of lymphodepletion and IL-2 (Table 1).

Table 2.

Characteristics of adoptive cell therapy studies including gynecologic cancer patients

Patient and clinical characteristics	Study arms (N = 32)	Gynecologic patients (N = 238)
Gynecologic patients, no. (%)		238 (100.0)
Ovarian cancer		147 (61.8)
Cervical cancer		81 (34.0)
Endometrial cancer		7 (2.9)
Vaginal cancer		2 (0.8)
Vulvar cancer		1 (0.4)
Median lines of prior therapy ≤1, no. (%)	4 (19.0)	51 (28.8)
Median lines of prior therapy >1, no. (%)	17 (81.0)	126 (71.2)
Study characteristics
Type of ACT, no. (%)
Peripheral autologous T-cell transfer	10 (31.3)	88 (37.0)
TCR	7 (21.9)	26 (10.9)
TIL	7 (21.9)	87 (36.6)
CAR T	6 (18.8)	22 (9.2)
NK	2 (6.3)	15 (6.3)
Type of study, no. (%)
Phase I	16 (57.1)	110 (46.2)
Phase I/II	1 (3.6)	7 (2.9)
Phase II	4 (14.3)	96 (40.3)
Pilot study	3 (10.7)	21 (8.8)
Case report	4 (14.3)	4 (1.7)
Lymphodepletion used, no. (%)
Yes	17 (53.1)	132 (55.5)
No	15 (46.9)	106 (44.5)
If yes, type of regimen, no. (%)
Cyclophosphamide plus fludarabine	12 (70.6)a	110 (83.3)
Cyclophosphamide	5 (29.4)	22 (16.7)
Target antigen, if any, no. (%)
HPV16	5 (26.3)	39 (38.6)
Folate receptor	3 (15.8)	40 (39.6)
Mesothelin	4 (21.1)	8 (7.9)
PD-1	3 (15.8)b	1 (1.0)
NY-ESO1	2 (10.5)	3 (3.0)
MUC1	1 (5.3)	7 (6.9)
MAGE A3	1 (5.3)	3 (3.0)
Patients in each study arm, median (range)	9.5 (2–33)c
Gynecologic patients in each study arm, median (range)	7.0 (1–27)c

^aOne study arm also included whole body radiation

^bTwo study arms included bispecific targeting of PD-1 plus HPV16 or mesothelin

^cExcluding case reports

ACT, adoptive cell therapy; TCR, T-cell receptor-modified T cells; TIL, tumor-infiltrating lymphocytes; CAR T, chimeric antigen receptor T-cells NK, natural killer cells

Patient-level ORR and DCR data are summarized in Table 3 and visually represented in S3. For all gynecologic cancers, 8.1% of patients had CR, 18.2% had PR, 31.4% had SD, for a total ORR of 26.3% and a DCR of 57.6%. The median duration of response was 5.5 months (range 2 to ongoing response at 29 months). By cancer type, ORR/DCR were 33.3%/65.4% for cervical cancer, 22.6%/54.1% for ovarian cancer, and 22.2%/44.4% for other gynecologic cancers; however, these rates were not significantly different. Patients in studies reporting 0 or 1 median line of prior therapy had improved ORR and DCR (both p≤0.001) compared to those reporting 2 or more. The DCR for patients in studies reporting 2 or more median lines of therapy was 53.2%.

Table 3.

Pooled patient-level objective response rate (ORR) and disease control rate (DCR) for use of adoptive cell therapy in gynecologic cancer (N = 238 patients)

	N	ORR, %	p valuea	DCR, %	p valuea
All gynecologic cancerb	236	26.3		57.6
Cancer type			0.204		0.183
Cervical cancer	81	33.3		65.4
Ovarian cancer	146	22.6		54.1
Other gynecologic cancer	9	22.2		44.4
Median lines of prior therapy			<0.001 *		0.001 *
≤ 1	51	52.9		80.4
> 1	124	22.6		53.2
Type of ACT			0.001 *		<0.001 *
TIL	87	41.4		73.6
NK	15	26.7		80.0
Peripheral autologous T-cell	87	18.4		44.8
TCR	26	15.4		53.8
CAR T	21	9.5		33.3
Lymphodepletion			0.001 *		<0.001 *
Yes	132	34.8		72.0
No	104	15.4		39.4
Target antigenc
HPV16	39	20.5		41.0
Folate receptor	40	15.0		32.5
Mesothelin	8	25.0		87.5
NY-ESO 1	1	0		66.7
PD-1	14	42.9		85.7
MUC1	7	14.3		14.3
MAGE-A3	3	33.3		33.3

^*Statistically significant p<0.05

^aPearson chi square test

^bMedian duration of response 5.5 months, range 2–29 and ongoing

^cp values not calculated due to low sample sizes

ACT, adoptive cell therapy; TIL, tumor-infiltrating lymphocytes; NK, natural killer cells; TCR, T-cell receptor-modified T cells; CAR T, chimeric antigen receptor T-cells

ORR significantly differed by type of ACT: 41.4% for TIL, 26.7% for NK cell, 18.4% for peripheral autologous T-cell transfer, 15.4% for TCR, and 9.5% for CAR T (p=0.001). This difference persisted when we compared TIL to all other ACTs (41.4% vs 17.4%, p<0.001). Patients who received lymphodepletion prior to ACT had a significantly higher response rate than those who did not (ORR, 34.8% vs. 15.4%, p=0.001; DCR, 72.0% vs. 39.4%, p<0.001). After we controlled for type of gynecologic cancer and lymphodepletion in a multivariable logistic regression, TIL was significantly associated with an objective response with an odds ratio of 2.6 (p=0.011, 95% CI 1.3–5.5, Table 4). Response also varied according to specific target antigens, with the highest ORR for therapies targeting PD-1, MAGE-A3, and HPV16, whereas the highest DCR was seen for therapies targeting mesothelin, PD-1, and NY-ESO1; however, the small sample size limits further analysis. Compared to non-gynecologic solid tumors that were included in these studies, ORR was not different for gynecologic cancers (26.3% vs. 27.8% non-gynecologic cancers, p=0.739).

Table 4.

Multivariable model of factors predictive of objective response in gynecologic cancer patients receiving adoptive cell therapy

Factors	Odds ratio (95% confidence interval)	p value
TIL	2.6 (1.3, 5.5)	0.011 *
Lymphodepletion	1.9 (0.9, 4.2)	0.101
Cervical cancer	0.8 (0.4, 1.7)	0.608

^*statistically significant p<0.05

TIL, tumor-infiltrating lymphocyte therapy

The rate of grade 3 or 4 adverse events in all patients included in the trials was 46.0% with study-level median rate of 29.2% (range 0,100; Table 5). Cytokine release syndrome (CRS)-related symptoms included grade 1 or 2 fever (68.2%), grade 3 or 4 fever (2.1%), grade 1 or 2 hypotension (5.1%), grade 3 or 4 hypotension (16.3%), grade 1 or 2 dyspnea (8.7%), and grade 3 or 4 dyspnea (14.3%). The incidence of these symptoms for patients receiving IL-2 support was slightly higher (G1/2 fever 89.9%, G3/4 fever 1.6%; G1/2 hypotension 5.1%, G3/4 hypotension 16.3%; G1/2 dyspnea 9.9%, G3/4 dyspnea 13.5%). Four papers^19–22 specified the diagnosis of CRS in a total of 12 patients, 1 of whom had a grade 3 or higher presentation. None of the included papers reported immune effector cell-associated neurotoxicity syndrome, although 4 cases of grade 3 confusion or delirium were reported. Common toxicities included grade 3 or 4 anemia, thrombocytopenia, and leukopenia as well as grade 1 or 2 nausea/vomiting, diarrhea, fatigue, and anemia. Due to inconsistent reporting, these and other study-specific toxicities are compiled as reported in the source literature in the supplemental section (S4). Febrile neutropenia was reported relatively frequently (45 events). Serious adverse events included tumor lysis syndrome, passenger lymphocyte hemolysis syndrome, compartmental CRS in the pleural cavity, pneumonitis, deep vein thrombosis, and an unrelated grade 5 fungal sepsis (one event each).

Table 5.

Toxicities reported in adoptive cell therapy studies including gynecologic cancer patients

	Grade 1 or 2, no. (%)	Grade 3 or 4, no. (%)
Fever	107 (68.2)	4 (2.1)
Hypotension	2 (5.1)a	17 (16.3)
Dyspnea	9 (8.7)	25 (14.3)
CRS	11 (38.5)a	1 (3.8)a
Confusion/delirium	4 (14.8)a	4 (5.3)a
All grade 3 or 4 toxicity	46.0% (N=237)
	Study-level median 29.2% (range 0,100, N=17)
Dose limiting toxicity	5 (6.1)a

Missing data were excluded.

^aN < 100

Discussion

In our heavily pretreated population of patients with advanced or recurrent gynecologic cancer, ACT seemed to be a promising treatment modality with an overall ORR of 26.3% and a median duration of response of 5.5 months. Lymphodepletion improved ORR. TIL therapy had a significantly superior ORR of 41.4%, which persisted after we controlled for cancer type and lymphodepletion. DCR in patients who underwent two or more median lines of prior therapy was 53.2%. Overall, ACT was well tolerated.

A major effort to improve the efficacy of ACT is focused on promoting the expansion and persistence of infused cells. Lymphodepletion is theorized to mediate the disruption of regulatory cells and immune tolerogenic mechanisms in the tumor microenvironment.²³ In a study in melanoma, the addition of lymphodepletion significantly improved both the proliferation of clonal T-cell populations and the persistence of this lymphocyte pool composition, which lasted over 4 months.²³ Clinically, this correlated to a more durable response to therapy.^13,14,24 In a Phase I study of CAR T-targeting folate receptors in ovarian cancer without lymphodepletion, there were barely detectable levels of infused cells 3 weeks post-treatment.²⁵ Similarly, Haas et al noted the expansion of CAR T cells targeting mesothelin in ovarian and other solid cancers was one tenth of that in hematologic cancers. Interestingly, addition of lymphodepletion improved expansion 10-fold but did not affect persistence.²⁶ In another study, when lymphodepletion by nonmyeloablative chemotherapy was augmented by addition of whole body radiation, there was no improvement in the expansion of NK cells.¹⁸ Clearly, the effect of lymphodepletion on the clonal cell population is not fully established; in fact, a study using TIL showed that clonal persistence may not predict clinical response at all.²⁷ In any case, our meta-analysis supports the use of lymphodepletion in improving oncologic outcomes.

Another strategy for improving in vivo expansion and persistence of ACT is suppression of inhibitory signaling, particularly through PD-1. Hirano et al demonstrated in murine models that constitutive and inducible PD-1 expression by tumors could confer evasion to therapeutic immunity.²⁸ Moreover, increased cytotoxic potency of ACT after PD-1 blockade was observed in preclinical studies.²⁹ In gynecologic cancers, high levels of exhaustion markers were noted in infused TILs, including LAG3 and PD-1, as well as MHC-II and PD-L1 expression in tumors.³⁰ Accordingly, several studies have incorporated PD-1 blockade. In a Phase I study of pembrolizumab-activated autologous T-cells for recurrent solid tumors, the one patient with gynecologic (ovarian) cancer had an SD.³¹ In a case report using CAR T encoding single-chain variable fragments (scFV) for mesothelin and antibodies for PD-1, the patient with ovarian cancer showed PR.³² More recently, O’Malley et al. reported on the second arm of cervical cancer patients who received LN145 TIL therapy in combination with pembrolizumab, which showed an astounding ORR of 50.0% in 10 paitents.³³ This was even higher than the ORR of 44.4% in arm 1 receiving LN145 monotherapy (27 patients).³⁴ The definitive contribution from PD-1 blockade is difficult to separate since arm 2 was also systemic-treatment naïve as opposed to the more heavily pretreated arm 1, and our data shows this as a predictor of response. In any case, the result is encouraging, and the scientific rationale is supportive. The mechanism of immune escape via stimulation of inhibitory receptors (PD-1, CTLA-4, and LAG3) is an area of active research.

Finally, the evolution of the tumor microenvironment with subsequent lines of therapy should be considered. While patients included in our meta-analysis had a median of 3 prior treatments, several studies conducted in the adjuvant or maintenance setting have shown improved survival benefit. In a study by Fujita et al, 24 patients who underwent cytoreduction followed by adjuvant chemotherapy (cisplatin, doxorubicin, and cyclophosphamide with or without fluorouracil) to no gross residual disease were non-randomly divided to either receive TIL therapy or undergo observation only. The addition of TIL improved 3-year progression-free survival (82.1% vs. 54.5%, p<0.05) as well as overall survival (100% vs. 67.5%, p<0.05).³⁵ Similarly, survival benefit was observed in biweekly or monthly maintenance autologous T-cell transfers after response to first-line therapy.^36,37 While some argue the earlier use of ACT leverages the short persistence of infused cells,³⁸ there is also data to support favorable immunomodulation by combined platinum-based chemotherapy,³⁹ particularly in a dose-dense regimen.⁴⁰ The result may be synergistic improvement of the effect of ACT. This may explain the unique success of Aoki et al. among the earlier studies as this protocol allowed upfront treatment with or without chemotherapy in addition to lymphodepletion.⁴¹ While our data suggests significantly higher response in patients with fewer prior lines of therapy, this conclusion is limited due to various confounding factors, and a head-to-head comparison is warranted.

A novel finding of this study is the remarkable success of TIL therapy compared to other ACT modalities, which persisted after controlling for lymphodepletion and the type of gynecologic cancer in a multivariable regression analysis. While the exact etiology is unclear, insight may be gained by examining the response to variously engineered HPV-targeting T cells, which provides an interesting equivalent comparison groups. A relatively robust ORR was reported for TIL targeting HPV (26.3% or 5 of 19 patients with gynecologic cancer)⁴² and TCR targeting HPV E7 (50.0% or 3 of 6 gynecologic patients).⁴³ In the Phase II study by Stevanovic et al, the magnitude of HPV reactivity of the infused TIL was associated with clinical response. However, only a small fraction of the polyclonal TIL contained oncoprotein-reactive T-cells (median 5%).⁴² Utilizing a personalized immunogenomic approach, the authors examined the two patients with cervical cancer who had a CR and found that both patients received TIL therapy that contained a relatively low frequency of HPV-targeted T-cells; in fact, 35% of the clonotype was targeting non-viral tumor neoantigens in the first patient, and 67% was targeting cancer germline antigens in the other patient.⁴⁴ It is possible that TIL therapy overcomes the issue of targeting and avidity in other ACTs by providing a broader spectrum of targets. In line with this, the authors of the recent TCR E7 study attributed the high response rate in part to the high-avidity peptide-HLA complex targeting and longer dissociation time.⁴³ Ongoing clinical trials include target antigens HPV16 E6/E7, NY-ESO1, MAGE-A3/A4, mesothelin, CD70, CD22, CD133, GD2, PSMA, MUC1, MUC16, HER-2, nectin4, anti-folate receptor alpha, ALPP, B7-H3, and TnMUC1.⁴⁵ In all, improvements in in vivo expansion/persistence, trafficking, and targeting of cells remain central to improving the efficacy of ACT.

There are several limitations of our study. Due to the limited amount of published data, heterogenous study protocols were included in this analysis, which constitutes the main limitation. While we saw significance of lymphodepletion in univariate analysis, this was not shown in multivariable regression (p=0.10), which likely represents limitations in sample size. Similarly, different subtypes of gynecologic cancers could not be analyzed separately due to their small sample size. This is particularly problematic for timing of therapy due to differing lines of standard of care options available for each subtype of cancer. The wide range of the study time periods may further add non-identifiable confounders over time. While gray literature articles allowed inclusion of cutting-edge data, this also led to missing variables and incomplete protocol information. Toxicity analysis was particularly limited due to the paucity of consistent patient-level data, which precluded in-depth calculations. Even for general rates presented herein, there was a degree of heterogenous reporting in “expected” events. Despite these limitations, this is the first pooled patient-level analysis of the use of ACT specifically in gynecologic cancer patients in current literature.

ACT has undergone vast therapeutic evolution over time. After the initial lack of success of peripheral autologous T cells, the field shifted towards TIL therapy with inclusion of improved conditioning regimen including lymphodepletion and IL2 support. TCR, CAR T, and NK cells are still in developmental nascency with room for improvement in efficacy. These, however, have the potential for more designer-specific targeting as well as off-the-shelf allogenic therapy not restricted by HLA compatibility. Strategies to improve in vivo cell expansion, persistence, trafficking, and antigen recognition should be pursued. From operational stand point, challenges remain in widely implementing this treatment modality, including the manufacturing process and scalability in the community. Innovations in streamlining the intensity of the regimen, infrastructure support, and recognition of unique toxicity and management are needed.

In all, ACT and especially TIL therapy hold promise in the care of advanced or recurrent gynecologic cancer. We suggest including lymphodepletion in the treatment regimen.

Supplementary Material

1

Supplemental 1. Full search strings for all databases

2

Supplemental 2. PRISMA 2020 flow diagram of search and selection process⁶⁰

4

Supplemental 4. Common toxicities reported in ACT studies including gynecologic cancer patients (all events)

3

Supplemental 3. Objective response rate (ORR) in patients with gynecologic cancer receiving adoptive cell therapy by (a) median lines of prior therapy, (b) type of ACT, (c) tumor-infiltrating lymphocyte (TIL) therapy vs. other ACT, and (d) whether lymphodepletion was used. (N = 238)

Highlights.

• ACT is a promising treatment modality in gynecologic cancer, with an ORR of 26.3% overall and up to 41.4% for TIL therapy

• TIL therapy was associated with objective response after controlling for tumor type and lymphodepletion

• The rate of objective response was higher when lymphodepletion was included in the treatment regimen

• The rate of grade 3 or 4 toxicity was 46.0%