A non-inferiority, phase III trial of gemcitabine plus capecitabine versus gemcitabine plus carboplatin as first-line therapy and tumor-infiltrating lymphocytes as a prognostic biomarker in patients with advanced triple-negative breast cancer

Abstract

Background:

Gemcitabine plus capecitabine (GX) shows survival benefit and manageable safety in patients with advanced triple-negative breast cancer (TNBC) but there is a paucity of phase III trial evidence. We aimed to compare the efficacy and safety of GX with gemcitabine plus carboplatin (GC) as first-line treatment for patients with advanced TNBC and validate the prognostic value of tumor-infiltrating lymphocytes (TILs).

Methods:

Patients with advanced TNBC were randomly assigned 1:1 to receive gemcitabine (1000 mg/m²) on days 1 and 8 plus oral capecitabine (1000 mg/m² twice a day) on days 1–14, or gemcitabine (1000 mg/m²) on days 1 and 8 plus carboplatin area under curve 2 on days 1 and 8. The primary endpoint was progression-free survival (PFS). TILs were analyzed by immunohistochemistry. The margin used to establish non-inferiority was 1.2.

Results:

In all, 187 patients were randomly assigned, with 93 in GX and 94 in GC. Median PFS was 6.1 months in the GX arm compared with 6.3 months in the GC arm. The hazard ratio for PFS was 1.148, and a 95% CI was 0.856–1.539, exceeding the non-inferiority margin of 1.2. The median overall survival (OS) was 21.0 months in the GX arm compared with 21.5 months in the GC arm. The safety profile for the GX regimen was superior to the GC regimen, especially regarding hematological toxicity. Patients with high CD8⁺ TILs had significantly longer PFS and OS compared with patients with low CD8⁺ TILs. In the high CD8⁺ TIL group, the GC arm had prolonged PFS and OS compared with the GX arm.

Conclusion:

The trial did not meet the prespecified criteria for the primary endpoint of PFS in patients with advanced TNBC. Moreover, the GC regimen showed better efficacy compared with the GX regimen in patients with high CD8⁺ TILs. However, the GX regimen should be considered in patients who cannot tolerate hematological toxicity.

Trial registration:

ClinicalTrials.gov identifier: NCT02207335.

Introduction

Triple-negative breast cancer (TNBC), characterized by the absence of estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor-2 (HER-2) overexpression, accounts for about 15–20% of all invasive breast cancers, has a more aggressive phenotype, lacks effective targeted therapy, and has a poor prognosis.^1–4 Due to the lack of routine endocrine and targeted therapies, chemotherapy remains the mainstay of the treatment of TNBC.^3,5 Due to the extensive utilization of combination chemotherapy based on taxanes and/or anthracyclines for early-stage breast cancer, the options for effective chemotherapy regimens for advanced TNBC are limited.^5,6

The efficacy of gemcitabine in treating advanced TNBC ranges from 15% to 28%.^7,8 The gemcitabine plus carboplatin (GC) regimen is the recommended standard regimen for advanced TNBC according to the practice guidelines of the National Comprehensive Cancer Network. However, the GC regimen exhibits severe treatment-related adverse events (TRAEs), especially myelosuppression and gastrointestinal toxicity. Capecitabine, an orally administered precursor of fluorouracil, demonstrates effectiveness in treating advanced breast cancer. Compared to other cytotoxic agents, capecitabine is a preferable option for physicians and patients due to its low toxicity and oral administration.^9,10 Several clinical trials have indicated that the gemcitabine plus capecitabine (GX) regimen may provide survival benefits for patients with advanced TNBC.^11–14 However, there is still a lack of phase III clinical trial evidence and treatment guidelines for the GX regimen in the first-line treatment for advanced TNBC.

The tumor microenvironment, a crucial mediator of cancer progression, plays a significant role in clinical prognosis and treatment response. ¹⁵ TNBC is the most immunogenic subtype among breast cancers, as evidenced by high levels of tumor-infiltrating lymphocytes (TILs) and programmed death-ligand 1 (PD-L1) expression in the tumor microenvironment. Levels of TILs have been proven to be associated with survival in TNBC patients. ¹⁶ Nonetheless, the prognostic and predictive values of TILs in advanced TNBC treated with different chemotherapy regimens have not been reported. We designed this phase III clinical trial to compare the efficacy and safety of GX with GC regimens in patients with advanced TNBC and explore the prognostic value of TILs.

Patients and methods

Study design

This multicenter, randomized, open-label, non-inferiority, phase III trial enrolled patients at two hospitals in China (Tianjin Medical University Cancer Institute and Hospital, and Cancer Hospital Chinese Academy of Medical Sciences). The study protocol was approved by medical ethics committee of Tianjin Medical University Cancer Institute and Hospital (E2013134), and the study was conducted in accordance with the Declaration of Helsinki, Consolidated Standards of Reporting Trials statement (CONSORT) and extension of the CONSORT 2010 statement for non-inferiority and equivalence randomised trials. ¹⁷ All patients provided written informed consent before enrollment. This study was registered on Clinical Trials.gov, number NCT02207335.

Patients

Patients who had not previously received systemic therapy for advanced TNBC were included. Other eligibility criteria included the following: female patients aged 18–75 years; diagnosis with negative ER, PR, and HER-2 phenotype (ER, PR, and HER-2 status was independently confirmed by two independent pathologists based on immunochemical analysis and in situ hybridization, and we used <10% positively stained cells as the cutoff value for ER/PR negativity in immunohistochemistry (IHC) testing according to the American Society of Clinical Oncology/College of American Pathologists guideline); Eastern Cooperative Oncology Group performance status (ECOG PS) of grade 0–2; measurable disease by CT or MRI according to Response Evaluation Criteria in Solid Tumors (RECIST, version 1.1); at least 6 months from the last adjuvant chemotherapy to recurrence or metastasis; and expected survival ⩾12 weeks.

Procedures

Female patients with advanced TNBC were randomly assigned (1:1) to receive gemcitabine (1,000 mg/m²) on days 1 and 8 plus oral capecitabine (1000 mg/m² twice a day) on days 1–14 every 21 days, or gemcitabine (1,000 mg/m²) on days 1 and 8 plus carboplatin area under curve 2 on days 1 and 8 every 21 days. After 6–8 cycles, patients continued to receive gemcitabine as maintenance therapy until progression.

The primary endpoint was progression-free survival (PFS). Secondary endpoints included objective response rate (ORR), clinical benefit rate (CBR), overall survival (OS), and safety. PFS was defined as the time from randomization to the first evidence of progression or death, and OS was defined as the time from randomization to death from any cause. ORR was defined as the percentage of patients who achieved complete response (CR) or partial response (PR). CBR was defined as the percentage of patients who achieved CR, PR, or stable disease (SD).

Dose delay and reduction were allowed. Gemcitabine and carboplatin doses could only be reduced twice to 75% and 50% of the initial dose, respectively. The capecitabine dose was allowed to be reduced twice to 80% and 60% of the initial dose, respectively. Gemcitabine, carboplatin, and capecitabine dosages were reduced for grade 3/4 febrile neutropenia, grade 4 neutropenia lasted for ⩾7 days, and grade 3/4 non-hematologic toxicities. In addition, the gemcitabine dose was reduced for grade 3/4 thrombocytopenia or thrombocytopenia-related bleeding.

Radiologic assessments were performed at baseline and repeated after every second cycle until disease progression using the same method (such as CT or MRI). Responses were evaluated according to RECIST version 1.1. Adverse events were recorded at every study visit and graded according to the National Cancer Institute Common Toxicity Criteria version 4.0.

Immunohistochemical staining method

As stromal TILs were the best parameter for the characterization of TILs, we selected stromal TILs to evaluate the relationship between TILs and the prognosis in advanced TNBC. ¹⁸ Antibodies were used to identify the infiltration of different types of TILs. CD3, CD4, CD8, and CD19 antibodies were purchased from Roche Co. Ltd. Immunohistochemical staining was performed on the Roche Benchmark XT automatic IHC instrument. CD3-, CD4-, CD8-, and CD19-stained cell membranes were brown. Stromal TILs were counted and confirmed by two investigators. Positive cells were counted microscopically in 5 fields (tumor beds) using a high-power objective lens (+40), and the counts were averaged. CD3 staining identified all T lymphocytes and CD19 staining showed all B lymphocytes. CD4 staining primarily marked the helper T cells, and CD8 staining showed the cytotoxic T cells. High infiltration was defined as positive TILs above the median, and low infiltration was defined as positive TILs below the median.

Multiplexed immunofluorescence detection

The multiplexed immunofluorescence staining was conducted at Genecast Biotechnology Co., Ltd. The slides with 4 μm thick sections were deparaffinized, rehydrated, and subjected to epitope retrieval by boiling in Tris-EDTA buffer for 15 min at 100°C, and then incubated in Antibody Diluent/Block (PerkinElmer #72424205) for 15 min. Primary antibodies for CD4 (ZM-0519, ZSGB-BIO, 1:100), CD8 antibody (ZA-0508, ZSGB-BIO, 1:100), and Foxp3 (ab20034, Abcam, 1:100) were incubated for 1 h at room temperature, and IDO (D5J4E, CST, 1:400) was incubated for overnight at 4°C. Anti-rabbit/mouse horseradish peroxidase antibodies (ZSGB-BIO # PV-6002 and PV-6001) were used as the secondary antibody and incubated at 37°C for 10 min. TSA visualization was then performed with the opal multiplex IHC Kit (#811001KT, PerkinElmer), containing fluorophores (DAPI), Opal 520 (CD4), Opal 620 (CD8), Opal 690 (Foxp3), Opal 570 (IDO), and TSA Coumarin system. Microwave treatment was performed with Tris-EDTA buffer for 15 min at 100°C. All slides were counterstained with 4′,6-diamidino-2-phenylindole for 5 min and enclosed in an Antifade Mounting Medium. Slides were scanned using the PerkinElmer Vectra. Multispectral images were unmixed with spectral libraries built from single stained tissue images for each antigen, using the inForm Advanced Image Analysis software (inForm 2.3.0; PerkinElmer, Waltham, MA, USA).

Statistical methods

This study employed a hybrid design to test the non-inferiority hypothesis. It was designed as a 1:1 randomized trial with a two-sided significance level of 0.05 and a power of 80%. The median PFS rates of GX and GC were assumed to be 5.4 months and 4.2 months, respectively, with enrollment for 48 months and follow-up for 12 months. The margin used to establish non-inferiority was 1.2. Accounting for a 5% dropout rate, 186 patients were required, with 93 patients per group.

The intention to treat (ITT) set was selected to analyze the efficacy and safety. All statistics were double-sided tests, with significant differences set at 0.05. Descriptive statistics were used to analyze the demographic characteristics. PFS and OS were analyzed with the Kaplan–Meier method and were compared between the two arms using the log-rank test. The Cox proportional hazard model was used to estimate the hazard ratio (HR) and corresponding 95% CI. The Pearson chi-square test was used to compare the ORR and CBR between the two arms. All who received at least one dose of chemotherapy were evaluated for safety. Safety analyses evaluated the rates of adverse events between treatment arms and were primarily descriptive. We used SPSS (version 23.0 for Windows, IBM, Chicago, IL, USA) and GraphPad Prism 8.0.1 (GraphPad Software, San Diego, CA, USA) to analyze all statistical data.

Results

Patient disposition

From January 2014 to December 2020, 203 female patients were enrolled. In all, 16 patients did not meet the eligibility criteria. The remaining 187 patients were eligible and randomly assigned, with 93 in GX and 94 in GC (ITT population for efficacy analysis, Figure 1). Every eligible patient received one or more cycles and underwent a safety assessment.

Figure 1.

CONSORT diagram.

CONSORT, Consolidated Standards of Reporting Trials statement; GC, gemcitabine and carboplatin; GX, gemcitabine and capecitabine; ITT, intention to treat; PD, progressive disease.

Patient characteristics and dose administration

Baseline characteristics and demographic indicators were balanced between the two arms (Table 1). The median age of the ITT population was 53 years (range, 29–75 years). 53.5% of patients had visceral metastases at baseline, and 47.1% had two or more metastatic sites. All patients received an average of 5.04 cycles, with 5.07 cycles in the GX arm and 4.99 cycles in the GC arm. In the GX arm, 36.6% of patients experienced dose reduction, and 8.6% experienced discontinuation, which was lower than that in the GC arm (51.1% and 13.8%, respectively).

Table 1.

Patient baseline characteristics and prior therapy.

Characteristic	GX (n = 93)	GC (n = 94)
Median age, years (Range)	53 (30–73)	52 (29–75)
ECOG PS
0	60 (64.5%)	63 (67.0%)
1	31 (33.3%)	29 (30.9%)
2	2 (2.2%)	2 (2.1%)
Type metastatic site
Visceral with or without other	49 (52.7%)	51 (54.3%)
Non-visceral only	44 (47.3%)	43 (45.7%)
Number of metastatic organ sites
1	48 (51.6%)	51 (54.3%)
2	24 (25.8%)	25 (26.6%)
⩾3	21 (22.6%)	18 (19.1%)
Metastatic sites
Lymph nodes	49 (52.7%)	52 (55.3%)
Chest wall or skin	21 (22.6%)	23 (24.5%)
Bone	24 (25.8%)	16 (17.0%)
Lung	34 (36.6%)	40 (42.6%)
Liver	17 (18.3%)	16 (17.0%)
Brain	4 (4.3%)	1 (1.1%)
Contralateral breast	4 (4.3%)	1 (1.1%)
Pleura	7 (7.5%)	5 (5.3%)
Pathological type
Ductal	84 (90.3%)	86 (91.5%)
Other	7 (7.5%)	6 (6.4)
Unknown	2 (2.2%)	2 (2.1%)
Hormonal receptor status
Both ER and PR < 1% positive	87 (93.5%)	82 (87.2%)
Others*	6 (6.5%)	12 (12.8%)
Time from surgery to recurrence
Primary metastatic	4 (4.3%)	2 (2.1%)
<12 months	20 (21.5%)	21 (22.3%)
⩾12 months	69 (74.2%)	71 (75.5%)
Neoadjuvant or adjuvant chemotherapy
Ye	87 (93.5%)	89 (94.7%)
No	6 (6.5%)	5 (5.3%)
Previous treatment with anthracycline
Yes	81 (87.1%)	83 (88.3%)
No	12(12.9%)	11 (11.7%)
Previous treatment with taxane
Yes	79 (84.9%)	83 (88.3%)
No	14 (15.1%)	11 (11.7%)

^*Others included ER < 1% positive and PR 1–9% positive, ER 1–9% positive and PR < 1% positive, and ER 1–9% positive and PR 1–9% positive.

ECOG PS, Eastern Cooperative Oncology Group performance status; ER, estrogen receptor; GC, gemcitabine and carboplatin; GX, gemcitabine and capecitabine; PR, progesterone receptor.

Efficacy

The ORRs in the GX arm and GC arm were 37.6% and 39.4%, respectively (p = 0.808; Table 2). The CBRs in the GX arm and GC arm were 78.5% and 79.8%, respectively (p = 0.828; Table 2). The median PFS of the ITT population was 6.3 months. One patient in the GX arm and four patients in the GC arm were censored. The Median PFS was 6.1 months in the GX arm and 6.3 months in the GC arm. The HR for PFS was 1.148, and a 95% CI was 0.856–1.539, exceeding the non-inferiority margin of 1.2 (Figures 2 and 3). Therefore, the trial did not meet the prespecified criteria for the primary endpoint of PFS. Several baseline covariates showed significant prognostic value on PFS, including time from surgery to recurrence, site of metastasis, and number of metastatic sites (Supplemental Table 1). At the time of the final OS analysis (140 deaths), the median OS for the ITT population was 21.0 months. In all, 21 patients in the GX arm and 26 patients in the GC arm were censored. The median OS was 21.0 months in the GX arm versus 21.5 months in the GC arm (HR, 1.002; 95% CI: 0.717–1.400; p = 0.992). Several baseline covariates showed significant prognostic value on OS, including ECOG PS, time from surgery to recurrence, site of metastasis, and number of metastatic sites (Supplemental Table 2).

Table 2.

Efficacy of study treatment.

Endpoint	GX (n = 93)	GC (n = 94)
Response
CR	2 (2.2%)	3(3.2%)
PR	33 (35.5%)	34(36.2%)
SD	38 (40.9%)	38(40.4%)
PD	20 (21.5%)	19(20.2%)
ORR	35 (37.6%)	37(39.4%)
p	0.808
CBR	73 (78.5%)	75 (79.8%)
p	0.828
PFS
Number of events	92	89
Median, months	6.1	6.3
95% CI for median	5.1–7.1	5.2–7.4
Log-rank test p	0.348
OS
Number of events	72	68
Median, months	21.0	21.5
95% CI for median	18.5–23.5	18.4–24.6
Log rank test p	0.992

CBR, clinical benefit rate; CR, complete response; GC, gemcitabine plus carboplatin; GX, gemcitabine plus capecitabine; PFS, progression-free survival; ORR, objective response rate; OS, overall survival; PD, progression of disease; PR, partial response; SD, stable disease.

Figure 2.

Kaplan–Meier plot of progression-free survival and overall survival in the ITT population.

GC, gemcitabine plus carboplatin; GX, gemcitabine plus capecitabine; HR, hazard ratio; ITT, intention to treat.

Figure 3.

Progression-free survival: univariable analysis by subgroup (Forest plot).

GC, gemcitabine plus carboplatin; GX, gemcitabine plus capecitabine; ECOG PS, Eastern Cooperative Oncology Group performance status.

Table 3 summarizes detailed information about anticancer drug treatment after the study. In all, 142 patients received other anticancer drug treatments, with 70 patients in the GX arm and 72 patients in the GC arm. The types of anticancer drugs were similar between the two arms. The most commonly prescribed drugs in the GX arm included carboplatin, vinorelbine, paclitaxel, cisplatin, and docetaxel. Similarly, the most frequently prescribed drugs in the GC arm included capecitabine, vinorelbine, paclitaxel, cisplatin, and docetaxel.

Table 3.

Summary of post-study anticancer drug treatment.

Variable	GX (n = 93)	GC (n = 94)
	No. %	No. %
Extent of chemotherapy, lines
None	23	22
Any	70	72
1	33	37
2	17	18
⩾3	20	17
No.	144	137
Single agent	52	58
Combination	92	79
Type of anticancer drug
Vinorelbine	31	24
Paclitaxel	31	21
Capecitabine	3	49
Carboplatin	34	3
Cisplatin	21	15
Docetaxel	21	16
Nab-paclitaxel	18	19
Pemetrexed	11	8
Bevacizumab	10	14
Apatinib	8	7

GC, gemcitabine plus carboplatin; GX, gemcitabine plus capecitabine.

Safety

Adverse events of any grade that the investigators judged to be TRAEs were recorded in all patients in both groups. Table 4 shows TRAEs with an incidence of at least 5% or TRAEs of grade 3/4. No treatment-related deaths were reported. All 187 patients had at least one adverse event. The incidence of TRAEs was numerically higher in the GC arm compared with the GX arm. The most common TRAEs of grade 3/4 in the GX arm included leukopenia (30%), neutropenia (24%), thrombocytopenia (16%), and hand–foot syndrome (16%). In the GC arm, the most common TRAEs of grade 3/4 included leukopenia (45%), neutropenia (51%), thrombocytopenia (29%), febrile neutropenia (12%), and anemia (12%). Therefore, the safety profile for the GX regimen was better than the GC regimen, especially in hematological toxicity.

Table 4.

Treatment-related adverse events.

Treatment-related adverse events	GX (n = 93)			GC (n = 94)
	Grade 1–2	Grade 3	Grade 4	Grade 1–2	Grade 3	Grade 4
Hematological AEs
Leukopenia	40 (43%)	24 (26%)	4 (4%)	34 (37%)	29 (31%)	13 (14%)
Neutropenia	34 (37%)	17 (18%)	6 (6%)	35 (37%)	28 (30%)	20 (21%)
Febrile neutropenia	–	2 (2%)	0 (%)	–	9 (10%)	2 (2%)
Anemia	33 (35%)	4 (4%)	0 (%)	46 (49%)	10 (11%)	1 (1%)
Thrombocytopenia	33 (35%)	13 (14%)	2 (2%)	28 (30%)	22 (23%)	6 (6%)
Non-hematological AEs
Hand–foot syndrome	41 (44%)	15 (16%)	–	4 (4%)	0 (%)	–
Alopecia	14 (15%)	–	–	60 (64%)	–	–
Diarrhea	25 (27%)	7 (8%)	0(%)	22 (23%)	5 (5%)	0 (%)
Nausea	25 (27%)	0 (0%)	–	51 (54%)	2 (2%)	–
Vomiting	12 (13%)	0 (0%)	0 (%)	27 (29%)	4 (4%)	0 (%)
Fatigue	23 (25%)	–	–	34 (36%)	–	–
Decreased appetite	15 (16%)	0 (0%)	0 (%)	25 (27%)	3 (3%)	0 (0%)
Peripheral sensory neuropathy	21 (23%)	4 (4%)	0 (%)	13 (14%)	0 (0%)	0 (0%)
Musculoskeletal pain	6 (6%)	0 (0%)	–	6 (6%)	0 (%)	–
Constipation	15 (16%)	0 (0%)	0 (0%)	15(16%)	1 (1%)	0 (%)
Pyrexia	8 (9%)	0 (0%)	0 (0%)	10 (11%)	0 (%)	0 (0%)
Infusion-related reaction	2 (2%)	0 (0%)	0 (0%)	10 (11%)	2 (2%)	0 (0%)
Increased ALT	9 (10%)	0 (0%)	0 (0%)	10 (11%)	0 (0%)	0 (0%)
Increased AST	15 (16%)	0 (0%)	0 (0%)	13 (14%)	0 (0%)	0 (0%)
Hyperglycemia	18 (19%)	1 (1%)	0 (0%)	24 (26%)	0 (0%)	0 (0%)
Hypocalcemia	7 (8%)	0 (0%)	0 (0%)	4 (4%)	0 (0%)	0 (0%)

AEs, adverse events; ALT, alanine aminotransferase; AST, aspartate aminotransferase; GC, gemcitabine plus carboplatin; GX, gemcitabine plus capecitabine.

The prognostic value of TILs in advanced TNBC

IHC was performed with antibodies against CD3, CD19, CD4, and CD8 antigens on tissue sections of 52 patients [Figure 4(a)]. Multiplex immunofluorescence staining was performed for CD4, CD8, FOXP3, and IDO1 on tissue sections of four patients [Figure 4(b)]. A higher infiltration of TILs was observed in the stroma compared with the intra-tumor. Multiplex immunofluorescence staining showed high infiltration of CD8⁺ and CD4⁺ TILs in the stroma compared with FOXP3⁺ and IDO1⁺ TILs. Patients with high CD8⁺ TILs were associated with prolonged PFS (HR, 0.559; 95% CI: 0.314–0.993, p = 0.047) and OS (HR, 0.436; 95% CI: 0.226–0.843, p = 0.014) versus patients with low CD8⁺ TILs (Figure 5). The infiltration of CD3⁺, CD4⁺, and CD19⁺ stromal TILs did not show prognostic value in our study (Figure 5). In the high CD8⁺ TIL group, patients treated with GC had prolonged PFS (HR, 0.322; 95% CI: 0.127–0.816, p = 0.017) and OS (HR, 0.300; 95% CI: 0.094–0.957, p = 0.042) compared with GX (Figure 6). In the low CD8⁺ TIL group, there was no significant difference in prognosis between the GC arm and GX arm (Figure 6).

Figure 4.

Tumor-infiltrating lymphocytes in invasive breast cancer. (a) Immunohistochemical detection of CD3, CD19, CD4, and CD8 (hpf, ×40); (b) Multiplex immunofluorescence staining of CD4, CD8, FOXP3, and IDO1 (hpf, ×40).

Figure 5.

Progression-free survival and overall survival according to (a) CD3⁺, (b) CD19⁺, (c) CD4⁺, and (d) CD8⁺ tumor-infiltrating lymphocyte counts, respectively.

HR, hazard ratio.

Figure 6.

Progression-free survival and overall survival according to GX or GC in the (a) low and (b) high CD8⁺ tumor-infiltrating lymphocyte groups, respectively.

GC, gemcitabine plus carboplatin; GX, gemcitabine plus capecitabine; HR, hazard ratio.

Discussion

TNBC is associated with an aggressive phenotype, severe heterogeneity, visceral metastasis, and a poor prognosis. ⁶ Metastatic TNBC is still considered incurable. Recent studies have demonstrated an impressive rationale for the use of many therapeutic targets.^19–22 KEYNOTE-355 and IMpassion130 had validated the efficacy of programmed death 1 (PD-1) or PD-L1 inhibitors in first-line treatment of TNBC.^19,20 Patients with breast cancer susceptibility gene (BRCA) mutations received a PFS benefit (1–3 months) after receiving poly (ADP-ribose) polymerase inhibitors (PARPi) treatment.^21,23 In addition, antibody–drug conjugates sacituzumab govitecan demonstrated impressive response rates, PFS, and OS benefits in patients with advanced TNBC. ²² However, chemotherapy remains a fundamental treatment option for advanced TNBC. According to clinical guidelines, single-agent chemotherapy is recommended for most patients due to its safety. However, many patients still receive combination chemotherapy as a first-line treatment strategy in China. ²⁴ Chinese patients with breast cancer are younger and have a higher rate of visceral metastasis than those in western studies. ²⁴ Combination chemotherapy can produce a faster and more significant response rate than single-agent chemotherapy and is usually used for visceral threats or symptomatic diseases. ²⁵ Therefore, combination chemotherapy is often prescribed in clinical practice in China.

We designed this phase III clinical trial to explore the non-inferiority of the GX regimen versus the standard GC regimen in the first-line setting for advanced TNBC. Although our data did not establish the non-inferiority of the GX regimen compared to the GC regimen, patients treated with the GX regimen achieved similar efficacy and experienced milder adverse reactions. Considering the safety and compliance of the GX arm, we believe that advanced TNBC patients who cannot tolerate hematological toxicity may still consider receiving the GX regimen. Compared to other combination chemotherapy regimens, the capecitabine-based combination regimen has better compliance and lower side effects. In addition, capecitabine can be chosen as a long-term maintenance treatment after combination therapy. Pembrolizumab, a PD-1 inhibitor, has been approved for treating locally recurrent unresectable or metastatic TNBC in combination with chemotherapy. However, it is crucial to note that this combination therapy is specifically approved for patients whose tumors express PD-L1 with a combined positive score of 10 or above. ¹⁹ PARPi, such as Olaparib, are specifically recommended for patients with BRCA gene mutations, as these inhibitors target DNA repair mechanisms that are compromised in these individuals. ²³ However, it is important to note that not all patients with TNBC have BRCA mutations. The proportion of patients with BRCA mutations is relatively small, estimated to be around 5–10% of all TNBC cases. ²⁶ Therefore, the efficacy of PARPi may be limited to this subset of patients with specific genetic alterations. The antibody–drug conjugate sacituzumab govitecan has shown promising efficacy in treating patients with advanced TNBC, significantly prolonging survival time. ²² However, it has not yet obtained approval for first-line treatment indications. In addition, its high cost and absence of medical insurance coverage in China have made it unaffordable for most patients. In summary, when compared to other treatments for TNBC, the GX regimen has significant advantages. It offers convenience, good tolerability, affordability, suitability for maintenance therapy, and effectiveness against metastatic disease.

The tumor microenvironment plays a significant role during breast cancer development and progression, and in determining the therapeutic response. TNBC has a unique immune microenvironment and typically exhibits high levels of TILs, which may be associated with its high immunogenicity and stronger immune response. ²⁷ Studies have shown that high levels of TILs in tumor tissue of TNBC patients are associated with improved outcomes, including reduced disease recurrence and extended OS.^16,28,29 This suggests that TILs may serve as a potential prognostic indicator and provide a means for predicting prognosis and guiding treatment strategies in TNBC patients.

Most of the infiltrating lymphocytes in breast cancer are CD4⁺ TILs and CD8⁺ TILs. Studies have found that approximately 60% of TNBCs contain abundant CD8⁺ TILs, and the cytotoxic effect of CD8⁺ T cells is more substantial in hormone receptor-negative breast cancer.^30,31 The cytotoxic effect of CD8⁺ T lymphocytes may be related to abundance, activity, and distribution. ³² As a helper T cell, CD4⁺ T lymphocytes participate in the activation of immune cells such as CD8⁺ T cells by differentiating into various subtypes. ³³ High CD8⁺ TILs are always associated with a better prognosis. It has been reported that the presence of CD8⁺ T cells is related to a significant reduction in the relative mortality risk in breast cancer. ³⁴ Jamiyan demonstrated that TNBC patients with good prognosis were dominated by stromal CD8⁺ TILs in both relapse-free survival and OS. ³⁵ Silva reported that, for post-neoadjuvant chemotherapy markers, patients with high CD8⁺ TILs had longer event-free survival in TNBC. ³⁶ Thus, the greater dominance of CD8⁺ TILs in TNBC suggests a robust immune response has been triggered. ³⁷ Our results indicated that patients with high CD8⁺ TILs had significantly longer PFS and OS compared with patients with low CD8⁺ TILs. Furthermore, our data showed that in the high CD8⁺ TIL group, patients treated with the GC regimen had prolonged PFS and OS compared with the GX regimen. However, further research is necessary to comprehensively understand the role of TILs in TNBC and determine their prognostic value in clinical practice.

This trial has several limitations. First, we used less than 10% as the cutoff value for ER/PR negativity, which was different from the latest modification of the College of American Pathologists guidelines. The different cutoff values for hormone receptor negativity led to a different definition of TNBC in our study, which caused a potential bias. In our study, six patients in the GX arm and 12 patients in the GC arm did not meet the new definition of TNBC. Second, 48.7% (91/187) of patients in our study did not receive a re-biopsy of recurrent or metastatic lesions, so inconsistent molecular subtypes could not be ruled out. Finally, as our study was an investigator-initiated multicenter clinical trial, we could not provide a totally centralized assessment of response and PFS due to financial limitations.

Overall, this phase III trial did not meet the prespecified criteria for the primary endpoint of PFS in patients with advanced TNBC. However, considering the safety and compliance of capecitabine, patients who cannot tolerate hematological toxicity can still consider receiving the GX regimen. Patients with high CD8⁺ TILs showed better outcomes. Moreover, for patients with high CD8⁺ TILs, the GC regimen could be a more feasible first-line chemotherapy strategy compared to the GX regimen. Further investigation is required to thoroughly explore its potential.