Introduction

Triple-negative breast cancer (TNBC), which lacks expression of estrogen and progesterone receptors (ER and PgR) and human epidermal receptor 2 (HER2), is associated with a dismal prognosis despite responding remarkably well to anthracycline and taxane-based neoadjuvant chemotherapy (NAC). In particular, standard polychemotherapy results in pathologic complete response (pCR) in more than 20 % of patients [1, 2], and this response is considered a surrogate of increased survival compared with patients without a pCR [3]. In a meta-analysis of 12 NAC studies, TNBC was associated with a pooled pCR (the absence of invasive and in situ cancer in the breast and axilla) of 34 % [4], and pCR was associated with significantly improved event-free survival compared with no pCR in TNBC patients (HR = 0.24, P < 0.001). In this meta-analysis, the improvement in the pCR odds ratio, however, did not correlate with an improvement in event-free and overall survival (OS).

There is an interest in DNA-damaging agents such as platinum drugs (cisplatin [CDDP] and carboplatin [CBDCA]) in TNBC. This interest derives from the fact that almost all TNBCs belong to the molecular subgroup of basal-like BCs according to the Perou classification [5]. These tumors exhibit high-proliferation rates and are seldom associated with BRCA1 mutations. A high proportion of TN patients exhibit BRCA1 functional alterations (BRCAness-like status), implying that these tumors are highly sensitive to interstrand cross-linking agents like platinum salts. In a registry of 6,903 patients, 10 out of 12 patients with BRCA1 mutations obtained a pCR when treated with single-agent CDDP [6]. Similarly, among 28 TNBC patients treated with four cycles of neoadjuvant CDDP, 22 % achieved a pCR, including two BRCA-mutated patients [7].

To date, however, no randomized phase III study has evaluated whether the addition of platinum salts to standard NAC including anthracycline and taxanes is capable of improving treatment efficacy in these patients. In 2013, at least two phase II randomized trials instead confirmed the benefit of adding CBDCA to NAC in TNBC. In a CALGB trial, the addition of CBDCA at the AUC of 6 increased the pCR rates from 28 to 42 % in stage II-III BCs [8]. Similarly, in a German study, the introduction of weekly CBDCA (AUC 1.5–2) to non-pegylated liposomal doxorubicin, weekly paclitaxel, and bevacizumab boosted the pCR rate from 37.9 to 58.7 % [9].

Here, we present a meta-analysis evaluating the association of TN histology with pCR after platinum-based NAC for operable or locally advanced BC as well as the benefit of the addition of platinum agents to conventional NAC. Furthermore, the activity of platinum salts in TNBC compared with non-TNBC was calculated.

Methods

Search strategy and selection of studies

PubMed, the Web of Science, EMBASE, SCOPUS, and the Cochrane Register of Controlled Trials (CENTRAL) were searched for studies (including conference abstracts) evaluating the pCR after platinum-based NAC in TNBC from 1990 to December 20th, 2013. We used the medical subject heading terms (“Breast Neoplasms”[Mesh] AND ((“cisplatin”[MeSH Terms] OR “cisplatin”[All Fields]) OR (“carboplatin”[MeSH Terms] OR “carboplatin”[All Fields]) OR (“platinum”[MeSH Terms] OR “platinum”[All Fields])) AND ((“neoadjuvant therapy”[MeSH Terms] OR (“neoadjuvant”[All Fields] AND “therapy”[All Fields]) OR (“neoadjuvant therapy”[All Fields] OR “neoadjuvant”[All Fields]) OR preoperative[All Fields] OR primary[All Fields])) and limited the results to English language studies. Eligibility criteria included randomized or nonrandomized studies reporting the proportion of pCRs (both in breast and axilla; ypT0N0) in TNBCs (defined as BC with ER and PgR expression in <1 % of cells and HER2-negative status) treated with CDDP or CBDCA-containing NAC, possibly including a taxane and/or an anthracycline. Studies including less than 10 patients as well as studies evaluating high-dose chemotherapy, phase I studies, targeted therapies alone, unconventional combinations (e.g., combinations not approved for advanced or localized disease), or therapies involving platinum as single agent alone were excluded from this analysis. In addition, the reference lists of the retrieved articles were checked to identify additional relevant publications. The “Related Articles” function was also used to improve the search. The study selection, data extraction, and data entry were performed by 2 authors independently (FP and AC), and discrepancies between the two reviewers were resolved by discussion and consensus. The final results were reviewed by the senior investigator (SB).

Data extraction

The following information was extracted from each article: (1) basic information, including the year of publication and the first author’s name; (2) study information, including sample size, study design, number of TNBC, and non-TNBC patients; (3) treatment information, including neoadjuvant schedules and number of cycles; (4) outcomes of interest, such as the percentage or number of pCRs in the TNBC population and the percentage (or number) of pCRs in the control arms for randomized studies (NAC without platinum agents); (5) percentage (or number) of pCRs and the ORR in non-TNBC treated with platinum agents; and (6) overall clinical RR (ORR), rate or number of breast-conserving surgery (BCS) in the TNBC population, long-term DFS and OS in TNBC subgroup, and DFS and OS in TNBC patients who obtained a pCR with platinum-based NAC.

Statistical analysis

Both the pooled pCR rates in TNBCs treated with platinum-based NAC and the comparison of the pCR rates of platinum- versus nonplatinum-based NAC in TNBC patients (for randomized studies) were the primary endpoints. Secondary endpoints were the comparison of pCRs in TNBCs and non-TNBCs treated with platinum agents, ORR, rate of BCS, DFS, and OS (for both all treated patients and pCR-only population) in TNBCs treated with platinum-based NAC. The pCR and other comparisons in the TNBC and non-TNBC subgroups were calculated using the method for dichotomous data (assessment of risk ratio [RR]; 95 % CI). Both the fixed-effect model/Mantel–Haenszel method with minimal heterogeneity in the variables among studies and the DerSimonian–Laird method (random effects model) when there was significant heterogeneity were used [10]. The Cochran’s Q test, with a predefined significance P threshold of 0.1, was used to assess the statistical heterogeneity among the studies. The assumption of homogeneity was considered invalid for P values less than 0.1; in this case, summary estimates were reported from the random effect models. Subgroup analysis was performed according to the platinum agent (CDDP vs. CBDCA) and chemotherapy schedule (platinum + taxane vs. platinum + taxane + anthracycline) (Table 1).

Table 1 Characteristics of included trials
Full size table

Finally, potential publication biases for the primary endpoints were evaluated using funnel plots, which assessed the relative symmetry of the individual study estimates around the overall estimate, followed by the Begg’s and Egger’s tests. A two-tailed P value < 0.05 without adjustment for multiplicity was considered statistically significant. The leave-one-out procedure was also performed for the primary endpoint analysis. The “fail-safe N” was calculated, which is defined as the number of additional “negative” studies (studies in which the intervention effect was zero) required to increase the P value for the meta-analysis to above 0.05.

A two-tailed P value <0.05 was considered statistically significant, and the results of the meta-analysis were reported as classic forest plots (for the primary endpoints). All statistical analyses were performed using NCSS 2007 software (version 07.1.21 released June 1, 2011) and Comprehensive Meta-Analysis software (version 2.2.064; July 27, 2011).

Results

Overall, a total of 3,850 references were identified, and 3,794 studies were excluded by reading the titles and abstracts. The studies were read further to identify case reports, clinical reports that did not provide pCR rates, nonconventional single-agent regimens, and studies where data extraction according to the TN molecular subtype was not possible to exclude a total of 28 studies. Ultimately, 28 studies were selected with a total of 1,598 TNBC and 596 non-TNBC patients [8, 9, 11–36]. Six were randomized phase II trials, and 22 were prospective or retrospective single-arm studies. The number of patients with TN disease treated with platinum drugs ranged from 10 to 315 in each study. The platinum agent was CDDP in eight studies, CBDCA in 18 studies, and both agents in one series (in one study this data was not reported). The reference flow is presented in Fig. 1.

Fig. 1
figure 1

Selection of publications included in the meta-analysis

Full size image

Primary endpoint: pooled pCR and RR for the comparison of platinum- versus nonplatinum-based NAC in TNBC

Overall, the pooled weighted pCR rate in all TNBCs treated with platinum-based NAC (n = 28 studies) was 45 % (95 % CI 40–49.9 %; Fig. 2). If we consider the trials that included both a taxane and anthracycline (n = 13 studies), this rate rises to 48.4 % (95 % CI 40–56 %).

Fig. 2
figure 2

The pooled ORR for platinum-based neoadjuvant chemotherapy in triple negative breast cancer

Full size image

Eight trials included CDDP-containing NACs. In one trial, a definition of platinum-based NAC did not permit us to split the CDDP- versus CBDCA-treated patients. The pooled pCR rate was 41.9 % (95 % CI 32–51 %). In 18 trials, CBDCA was the platinum of choice and was associated with a pCR rate of 46.3 % (95 % CI 40.7–52.1 %).

Compared with nonplatinum-based NAC, the addition of a platinum agent in five randomized trials increased the pCR rate by 45 % (RR 1.45, 95 % CI 1.25–1.68, P < 0.0001; I 2 = 31.7 %, P for heterogeneity 0.2 according to fixed-effect model; Fig. 3). The pCR rate increased from 32 to 48 % (absolute increase 16 %, 95 % CI 10–22 %, P < 0.0001).

Fig. 3
figure 3

The forest plots of RR with 95 % CIs for pCR in triple-negative breast cancer treated with versus without platinum-based neoadjuvant chemotherapy

Full size image

pCR in TNBC versus non-TNBC with platinum-based NAC

Compared with non-TN histology, TNBCs were associated with a twofold increase in the rate of pCR when treated with platinum-based NAC in 13 studies (from 19.6 to 48.4 %; RR 3.32, 95 % CI 2.39–4.61, P < 0.0001, I 2 < 0.0001, P for heterogeneity 0.74 according to fixed-effect model).

ORR in TNBC versus non-TNBC, BCS rate, DFS, and OS in all and pCR-only patients

In all TNBC treated with platinum-based NAC, the pooled ORR was 86.7 % (95 % CI 82.7–89 %). This rate was similar to TNBCs treated without CDDP or CBDCA in two studies (P = 0.164).

The rate of breast-conserving surgeries ranged from 7.6 to 86 %. Long-term DFS and OS ranged from 50 to 76 % and from 49.5 to 89 %, respectively. In particular, DFS and OS in patients who achieved a pCR were excellent (range 81.8–90 % for DFS and 78–95.6 % for OS) and were superior to the DFS and OS of TNBC that did not obtain a pCR (range 44-56 % for DFS and 50–51 % for OS). The RRs of progression and death for pCR versus non-pCR patients were 0.29 (95 % CI 0.18–0.49, P < 0.0001) and 0.43 (95 % CI 0.24–0.75, P = 0.004), respectively.

Publication bias

A funnel plot and both the Begg’s and Egger’s tests were performed to assess the publication bias of the selected studies for the pooled pCR rate analysis and for comparison of platinum- versus nonplatinum-containing NAC. The shapes of the funnel plots revealed little evidence of asymmetry for pooled pCR analysis in TNBC treated with platinum-based NAC (Fig. 4). However, the Begg’s test (P = 0.7) and Egger’s test (P = 0.89) were not significant for pooled pCR analysis. The results remained unchanged after the leave-one-out procedure. Using the trim and fill method to account for asymmetric studies in the funnel plot, there was no effect on the pCR rate. The fail-safe N was 53, indicating that it would be necessary to locate and include 53 “null” studies for the combined 2-tailed P value to exceed 0.05.

Fig. 4
figure 4

Funnel plot for publication bias in pooled pCR analysis (triple-negative populations only)

Full size image

Similarly, for the comparison of platinum- versus nonplatinum-based NAC in TNBC, there was no evidence of asymmetry in the funnel plot. Both Begg’s (P = 0.46) and Egger’s tests (P = 0.44) were not significant. Using the trim and fill method to account for asymmetric studies in the funnel plot, there was no effect on the RR (RR 1.36, 95 % CI 1.07–1.71). The fail-safe N was 24, indicating that it would be necessary to locate and include 24 “null” studies for the combined 2-tailed P value to exceed 0.05.

Discussion

This systematic review and meta-analysis suggests that TNBC is associated with a high rate of pCR when treated with platinum-based NAC and that the addition of platinum salts increases this rate by 31–45 % in five randomized studies. OS and DFS data are not available to evaluate whether the addition of platinum agents to NAC improves TNBC outcomes, and larger trials with longer follow-ups are necessary. Furthermore, the TN subgroup is associated with a pCR rate nearly double that of non-TNBC when both are treated with platinum-based NAC, suggesting that TN status is a potential therapeutic target for these drugs.

Our review lead to the observation that the treatment of TNBC with platinum salts is associated with a near similar pCR rate as HER2 + BC treated with modern anti-HER2 drugs, such as pertuzumab and lapatinib, added to trastuzumab. The NeoSPHERE and NeoALLTO studies demonstrated infact that the double HER2 blockade similarly increased the pCR rate from 29 to 45 % with pertuzumab plus trastuzumab and from 29 to 51 % with lapatinib plus trastuzumab [37, 38]. In patients who achieved a pCR, there was also an increase in event-free survival and OS compared with non-pCR patients as demonstrated in the NeoALLTO study (HR 0.38 and 0.35). Among trials analyzed in the present review, only three trials provided long-term DFS and OS, and these outcomes were excellent for TNBC treated with platinum-based NAC. Furthermore, in patients who achieved a pCR, there was a significant benefit in relapse and death (70 and 57 % less risk of relapse and death) compared with non-pCR TNBC patients treated with the same platinum agents. In the Cortazar meta-analysis [4], TNBCs (as well as HER2 + and ER + grade 3 BCs) that achieve a pCR in breast and axilla (ypT0N0) were associated with a better event-free survival compared with patients with residual disease. These data and our meta-analysis suggest that adding an active agent to a backbone NAC with demonstrated efficacy could result in a meaningful significant increase of the pCR rate, and a potential cure for some TNBC patients.

Whether all TNBC patients would benefit from platinum agents during preoperative chemotherapy is a matter of debate. It is likely that BRCA-mutated tumors, representing up to 50 % of all TNBC [39, 40], are the most sensitive to DNA-damaging agents. In the Birsky and Silver publications cited above [6, 7], 83 and 100 % of BRCA-mutated patients obtained a pCR with single-agent CDDP. However, BRCAness tumors that shared mixed characteristics of sporadic cancer and inherited BRCA-mutations were more frequent (approximately two-thirds of all TNBCs) and were associated with decreased pCR compared with BRCA-mutated tumors (35 vs. 63 %) [41]. Similarly, BRCA-mutated TNBCs were associated with a better prognosis than nonmutated tumors [41]. In our series, only one trial reported the different responses of TNBC with and without BRCA mutation; these response were 100 and 15 %, respectively, confirming this suggestion. From a molecular point of view, TNBC is a group of different entities as demonstrated by Masuda et al. [42]. At least 7 subtypes were described, with basal-like-1 that was associated with the highest pCR rate (52 %). This notion is hugely crucial and needs to be validated prospectively; this could permit a better elucidation of responsiveness of TNBC to platinum agents.

Other agents have tested as neoadjuvant therapy in TNBC. The anti-VEGF-A monoclonal antibody bevacizumab was a candidate agent for use in aggressive TNBC subsets. Among more than 600 TNBC patients treated into the GeparQuinto trial and randomized to NAC with four cycles of anthracyclines followed by four cycles of docetaxel with or without bevacizumab added to NAC, the pCR rates (ypT0ypN0 after surgery) were 27.9 % without and 39.3 % with bevacizumab (P = 0.003), similar to our results [43]. In our series, trials including bevacizumab as a part of platinum-containing NAC were associated with a pCR rate of 52 %. In the recent CALGB trial lead by Sikov, the addition of bevacizumab to NAC with or without CBDCA increased the pCR by 10 % [9]. Another explored target in TNBC is EGFR. A randomized phase II trial in early BC with ER-negative disease explored the addition of gefitinib to anthracycline NAC. The pCR rate was increased by 5 % with the anti-EGFR agent compared with placebo, but the rate was nonetheless poor (17 vs. 12 %). However, a post-hoc subgroup analysis revealed a significant difference in pCR between TNBC and non-TNBC tumors (P = 0.03) [44]. The poly(ADP-ribose) polymerase (PARP) inhibitor veliparib, with chemosensitizing and antitumor activities, confers an added benefit when added to CBDCA and paclitaxel plus standard anthracycline-based NAC. In the randomized phase II study I-SPY 2, the researchers observed a pCR in 52 % of women who were treated with veliparib plus CBDCA and paclitaxel versus a 26 % pCR rate in those who received paclitaxel alone. Both regimens were added to conventional anthracyclines [25].

The best platinum salt to be added to NAC is presently unknown. All randomized trials included in our study used CBDCA-based NAC, suggesting that CBDCA should be applied in clinical practice. In eight studies, CDDP was the agent of choice and resulted in a pCR rate of 41.9 %, similar to the overall pooled analysis and slightly inferior to CBDCA-based NAC studies (46 %). In the Hurley series, however, the use of CDDP, but not CBDCA, was an independent predictor of PFS and OS [30].

Two issues remain unsolved: the prognosis of BC patients with TN histology and residual disease after NAC, and the prediction of patients more likely to obtain a pCR. This and other studies [1, 2] demonstrate that TNBC, which does not get a pCR after NAC, is associated with a dismal prognosis. Early identification and treatment with a non-cross-resistant agent could improve responses and prognoses. In GeparTrio study, however, a different schedule in early nonresponders after 2 cycles of NAC did not permit to obtain a better pCR rate or an improved DFS in n = 362 TNBC patients enrolled [45]. Molecular profiling of residual tumor burden in 74 TNBCs treated with NAC identified genetic alterations potentially treatable with targeted therapies in 90 % of cases [46]. Early nuclear medicine tools such as PET could also improve the detection of TNBC NAC non-responders [47, 48]. Other molecular predictors, including grade, young age, BCL-2 expression, p53, high Ki67, LDH, and basal-like status, were associated with better results with NAC in TNBC, but none of these predictors were analyzed in platinum-treated TNBC patients [49–54].

Our systematic review has some limitations. This is a meta-analysis of published trials, and only six studies included a randomized design. The included studies represent a mixed population of operable and locally advanced TNBCs with different prognoses and responses to NAC. Finally, the NAC schemes comprise conventional and nonconventional schedules (e.g., ECF schedule and gemcitabine/adriamycin + gemcitabine/CDDP combination) with slightly different durations. It is likely that polychemotherapy including both anthracyclines and taxanes may obtain the best results as previous publications have confirmed. Nonetheless, our meta-analysis represents about 1,600 TNBCs, all treated with platinum-based NAC. To our knowledge, the present study represents the largest review ever published about this controversial topic. Finally, comparing five randomized phase II trials, we are able to calculate a significant pCR rate in association with platinum salts plus both taxanes and anthracyclines.

This meta-analysis updated a previous Chinese review that included only three out of 28 trials analyzed here [55]. These results are similar to those presented in this review, although the previous authors did not include any randomized studies. The Liu et al. meta-analysis was furthermore extended to metastatic patients, in whom the addition of platinum agents increased ORR but did not improve PFS or OS. Their data confirm an older meta-analysis offered by the Cochrane Collaboration review, which demonstrated an ORR benefit but did not observe any survival or time to progression increase with platinum agents [56]. The different effects of platinum salts in the localized versus metastatic disease suggest that these agents should be introduced early in the course of the disease.

In conclusion, the optimal NAC combination/sequence incorporating platinum salts in the neoadjuvant setting is far from evident given the data available. Nevertheless, the best candidates for platinum-based NAC for BC seem to be those with the TNBC phenotype because these patients respond better compared with non-TN patients and achieve a significant improvement in pCR rates when CDDP or CBDCA is added.