Facts and Hopes in Immunotherapy for Early-Stage Triple-Negative Breast Cancer Open Access

Triple-negative breast cancer (TNBC) is characterized by the lack of expression of the estrogen receptor (ER) and progesterone receptor and no amplification of the HER2 receptor. Typically, TNBCs are poorly differentiated and have a high proliferation rate. Approximately 40% to 50% of early-stage (e)TNBC patients achieve a pathologic complete response (pCR) to neoadjuvant chemotherapy (NAC), which translates into excellent survival (1). Increased levels of stromal tumor-infiltrating lymphocytes (sTIL) are clearly associated with a high likelihood to reach a pCR after NAC in TNBC. Nevertheless, historically, breast tumors have long been considered as immunologically “cold” with a lower level of neoantigens and immune cell infiltration compared with other solid tumors (2, 3). Decades of research have proven there is immense heterogeneity between breast tumors (4) in terms of genetic drivers, immune infiltration (5–7), and treatment response. The immune infiltration of TNBC has particularly gained interest over the past decade. We now know that eTNBC has relatively high levels of TILs compared with, e.g., luminal breast cancer (5) and that the level of TILs in eTNBC may serve as a surrogate marker for the endogenous immune response in the tumor. This is exemplified by the positive correlation between the level of sTILs, prognosis (8), and response to chemotherapy in eTNBC patients. eTNBC patients, therefore, should be considered as having a potentially immunologically active tumor that can be exploited by therapies targeting activation or reinvigoration of immune cells. Chemotherapy partially exploits this inert immunity alongside its classic goal of direct tumor lysis, by inducing immunogenic cell death and by specific immunomodulatory capacities such as increased MHC class I expression or reduction of immunosuppressive cells (9).

Until recently, the standard of care (SOC) for eTNBC consisted of curative-intent systemic chemotherapy, radiotherapy, and surgery. Neoadjuvant chemotherapy is the cornerstone of current practice and the treatment landscape is rapidly changing with new (neo)adjuvant additions, such as carboplatin (10), capecitabine (11), or olaparib, for BRCA1/2 mutation carriers (12). Recently, the addition of immune-checkpoint inhibitors (ICI) to chemotherapy has been explored in several clinical trials. The addition of programmed death 1 (PD1) blockade to chemotherapy increases pCR rates and event-free survival (EFS) for eTNBC patients (13–15), proving that not all breast tumors are too cold to awaken. This milestone should be considered only as the starting point of immunotherapy implementation in daily clinical practice as major challenges lie ahead of us to achieve personalized medicine for eTNBC.

In this review, we will first briefly summarize where we stand now with immunotherapy in eTNBC. Next, the scientific challenges that lie ahead of us to optimize, time and combine immunotherapy for eTNBC will be discussed and how the use of baseline and dynamic biomarkers will potentially guide us in more precise patient and therapy selection.

The results of several phase II and III trials, combining ICI with NAC in eTNBC, have been reported over the last years (13, 15–19). The pivotal phase III KEYNOTE-522 trial (15) randomized stage II–III TNBC patients to neoadjuvant chemotherapy, including anthracyclines/cyclophosphamide (AC) plus taxanes/carboplatin with or without the addition of neoadjuvant and adjuvant pembrolizumab. The pCR rate in the placebo arm was 51.2% as compared with 64.8% in the pembrolizumab arm in the first 602 patients (Table 1), and 55.6% in the placebo arm compared with 63% in the pembrolizumab arm (15) at the second interim analysis with 1,174 patients randomized. With a hazard ratio of 0.63, a statistically significant improvement in EFS was seen in patients receiving neoadjuvant and adjuvant pembrolizumab compared with the placebo arm (ref. 15; Table 1). These intriguing results led to the FDA approval (20) and recent EMA recommendation (21) for this regimen for eTNBC. Recently, it was shown that particularly patients with a residual disease burden (RCB) category II (moderate residual disease) had improved EFS with the addition of pembrolizumab, illustrating that for this group of patients, ICI adds to long-term tumor control, but this effect was limited in patients with RCB category (22) 0 (corresponding with pCR), I (minimal residual disease), or III (extensive residual disease; ref. 23). It is tempting to speculate why there are differences in EFS according to RCB category. Patients with RCB 0 or I after NAC probably have an excellent prognosis without the addition of ICI, whereas patients with RCB III most likely comprise the subgroup of eTNBCs with aggressive features and limited immune cell infiltration nonresponsive to chemotherapy or ICI. It remains to be determined why patients with RCB II have long-term benefit from the addition of ICI, but we hypothesize that NAC plus ICI shapes the TME of this subgroup into an immune-rich system with long-term immune control.

Similarly, the combination of neoadjuvant atezolizumab and AC plus nab-paclitaxel, including 1 year of adjuvant atezolizumab, in the IMpassion-031 (17) significantly increased pCR rates from 41% in the placebo arm to 58% in the atezolizumab arm. Although mature EFS has yet to be reported, the first interim analysis showed a nonsignificant hazard ratio of 0.67 in favor of atezolizumab (Table 1).

Three other randomized phase II–III neoadjuvant clinical trials have reported their primary endpoint. In the I-SPY2 trial arm with pembrolizumab plus AC and paclitaxel, an estimated pCR rate was observed of 60%, whereas this was 47% in the arm with durvalumab and olaparib plus AC and paclitaxel. pCR rates in the rolling control arm were 22% and 27%, respectively, which is substantially lower as compared with historical trials (Table 1). In the GeparNuevo (16), the addition of neoadjuvant durvalumab to AC and nab-paclitaxel led to a numerical improvement of pCR rates, from 44.2% in the placebo arm to 53.4% in the durvalumab arm, but this was not statistically significant (P = 0.287). The 3-year disease-free survival (DFS) was numerically higher in the durvalumab arm (85.6%) than in the placebo arm (77.2%; P = 0.036), the 3-year distant (D)DFS 91.7% vs. 78.4% (P = 0.005), and 3-year overall survival 95.2% vs. 83.5% (P = 0.006; ref. 24). Durvalumab added to neoadjuvant chemotherapy in TNBC therefore improved long-term outcome despite a small pCR increase and no continuation of ICI after surgery (24). In the NeoTRIP study (25), atezolizumab was added to neoadjuvant carboplatin and taxanes, and in this study no numerical difference in pCR rate was observed between the placebo and the durvalumab arms; however, the primary endpoint of this study was EFS and has yet to be reported (Table 1). Additionally, in the single-arm NeoPACT study (19), a pCR rate of 58% was observed in TNBC patients treated with neoadjuvant pembrolizumab, docetaxel, and carboplatin with a 2-year EFS rate of 89% in all patients and 98% in the patients with a pCR without adjuvant use of pembrolizumab.

Although the comparison between trials is hampered by differences in patient selection, experimental drug, an adjuvant ICI component and chemotherapy backbone, it is tempting to speculate on the results seen so far in the neoadjuvant immunotherapy TNBC space. Although IMpassion-031 and KEYNOTE-522 report significantly higher pCR rates with the addition of ICI, this was not observed in the smaller trials NeoTRIP and GeparNuevo. One of the reasons could be the lack of anthracyclines in the chemotherapy backbone in NeoTRIP as previous studies suggested potential synergy of anthracyclines with ICI (26, 27). However, the pCR rate of 58% in the NeoPACT trial, which also omitted anthracyclines, would argue against this. Additionally, slight differences in baseline characteristics such as a high rate of node-positive patients in the NeoTRIP trial could have potentially negatively affected the clinical results. Moreover, it is still up to debate whether or not pCR is the best endpoint of interest for patients treated with chemotherapy plus ICI for eTNBC. As the GeparNuevo (24) trial had no adjuvant ICI component and still demonstrated a survival benefit of neoadjuvant durvalumab over placebo, it remains to be determined whether adjuvant ICI in all patients or patients with residual disease is needed. Three-year EFS rates in the KEYNOTE-522 with adjuvant ICI seem to be remarkably similar to 3-year DFS rates in the GeparNuevo without adjuvant ICI (85% in the experimental arms vs. 77% in the placebo arms). As data from other tumor types show that ICI is presumably more effective in the neoadjuvant setting with an in situ primary tumor needed to elicit an immune response (28, 29), it is likely that adjuvant ICI after neoadjuvant ICI does not contribute much to survival. Whether this is the case in TNBC is currently evaluated in ongoing clinical trials (NCT03281954, OptimICE-pCR), yet the expectation is that adjuvant ICI after pCR might be redundant and that adjuvant ICI after 5 months of neoadjuvant ICI in case of a non-pCR lacks sufficient efficacy.

Although anti–PD-1 (aPD-1) and anti–PD-L1 (aPD-L1) are rapidly changing the SOC for eTNBC, other immunotherapeuticals and new combinations of (immune)therapies are entering the (neo)adjuvant regimens of patients with eTNBC in clinical studies. In parallel with the uprise of ICI in TNBC, new therapeutic strategies that effectively target other key hallmarks of cancer (30) and with potential synergy to ICI are rapidly emerging. In addition, several studies focus on finding the optimal chemotherapy backbone to accompany ICI. This section discusses neoadjuvant and adjuvant upcoming combinatory therapies that could be considered.

Beyond anthracycline- and taxane-based neoadjuvant chemotherapy, new neoadjuvant combination therapies with ICI are considered for patients with eTNBC. Most studies on ICI in TNBC so far have focused on the PD-(L)1 axis, yet the combination of ICI agents, such as but not limited to aCTLA-4, is common in other solid tumor types (31–34). In the phase II DART trial, the combination of aPD-1 and aCTLA-4 was tested in patients with advanced metaplastic breast cancer with a response rate of 18% with all responses lasting for at least 2 years, indicating durable responses can be elicited by this combination therapy (35). In eTNBC, neoadjuvant aPD-1 and aCTLA-4 combined with paclitaxel in patients with stage 3 TNBC with residual disease after anthracycline-based chemotherapy resulted in a response rate of 58% and a pCR rate of 18% (36). Although these data are relatively disappointing, this poor patient population potentially needs additional or different treatment to overcome immune evasion. However, it is important to acknowledge that nonresponders to chemotherapy and ICI are a preselection of patients with low immunogenic tumors and with potentially other mechanisms of treatment resistance, indicating we may have to expand our treatment repertoire beyond these SOC therapies to kick-start an effective antitumor response. Moreover, the ongoing phase II BELLINI trial (37) evaluates the increase in CD8 T-cell and IFN-gamma response and radiologic response after aPD-1 with or without aCTLA-4 in the neoadjuvant setting with only 4 to 6 weeks of ICI, exploring if ICI combination therapy alone can induce immune activation and clinical responses (NCT03815890). It is expected that new combinations with aPD-1, targeting other immune-checkpoint such as LAG3 (38) and TIGIT (39), will be tested in phase II and III trials in the eTNBC field in the coming years. It is still unclear how the effect of aPD-1 could be enhanced by targeting checkpoints that overlap with aPD-1 in terms of expression and function and if we therefore can truly expect a gain in response rates from these combinations in eTNBC. The first results from advanced melanoma and lung cancer trials lead to muted expectations so far. aLAG3 is currently being tested in advanced and metastatic TNBC. In advanced and metastatic melanoma, aPD-1 plus aLAG3 resulted in a significant better PFS (38), yet no significant improvement in overall survival was reported (40). Anti-TIGIT recently failed in the SKYSCRAPER-02 trial (41), where anti-TIGIT was investigated in combination with atezolizumab and chemotherapy carboplatin and etoposide for the first-line treatment of extensive-stage small-cell lung cancer; the combination did not meet the primary endpoint of PFS. Moreover, we still have limited information about what the expected toxicity is of the combination of aLAG3 or aTIGIT when combined with neoadjuvant chemotherapy regimens for eTNBC. If and how novel immune-checkpoint inhibitors such as but not limited to aLAG3 and aTIGIT are introduced in the (neo)adjuvant space of eTNBC remains to be seen.

Beyond the combination of multiple ICI targets, ICI can be combined with new neoadjuvant agents targeting other hallmarks of cancer than immunosuppression. The ultimate goal for these combinatorial therapies is to turn a noninfiltrated tumor into a hot—immune responsive—tumor. PARP inhibitors have proven effective in metastatic breast cancer patients with a germline BRCA1 or BRCA2 mutation (42, 43) but also have the potential to be effective in cancers with other defects in homologous recombination DNA repair. Four weeks of neoadjuvant olaparib with one dose of durvalumab followed by neoadjuvant chemotherapy resulted in high pCR rates (75%) in TNBC (ER <10%) stage II/III breast cancer (44). Among 13 patients with germline BRCA mutations, 11 achieved pCR (84.6%; ref. 44). The value of PARP inhibition and ICI without chemotherapy still needs to be studied. Indeed, future clinical trials, as exemplified by the EORTC-sponsored phase II trial (NCT05209529), explore the combination of neoadjuvant PARP inhibition with or without ICI, without chemotherapy for stage I/II BRCA-mutated TNBC or HRD/BRCANess TNBC patients. Recently, antibody–drug conjugates (ADC) have been changing the therapeutic field of metastatic breast cancer, especially with the recent presentation of the Destiny-Breast-04 data, which demonstrated an overwhelming survival benefit with trastuzumab deruxtecan (T-Dxd) over SOC chemotherapy in patients with metastatic HER2-low breast cancer (45). In the phase III ASCENT study, PFS and overall survival were significantly longer with sacituzumab govitecan (SG), a Trop-2–directed ADC, than with single-agent chemotherapy among patients with metastatic TNBC (46). The role of SG in eTNBC and whether or not synergy exists in combining SG with ICI remains to be explored. SG is currently being evaluated in the NeoSTAR clinical trial (NCT04230109) with or without aPD-1 in the neoadjuvant setting for patients with localized TNBC. Preliminary data on neoadjuvant SG monotherapy showed that the radiologic response rate with SG alone was 62% after 4 cycles and that the pCR rate was 30% (47), indicating neoadjuvant efficacy in localized TNBC. The phase II BIS program (NCT05180006) evaluates short treatment of atezolizumab monotherapy with or without other biological agents, such as ipatasertib (AKT inhibitor) or bevacizumab (VEGF inhibitor), and studies levels of activated GzmB⁺ CD8⁺ T cells on treatment. However, preliminary results of the BARBICAN trial reported that the addition of ipatasertib to neoadjuvant atezolizumab plus chemotherapy did not improve pCR rates in patients with eTNBC (48). Talimogene laherparepvec (T-VEC) is an attenuated replication-competent herpes simplex virus type 1 (HSV-1) that lyses tumor cells and can additionally elicit an antitumor response by releasing tumor-associated antigens and providing cytokine stimuli (49). This can potentially lead to an enhanced tumor antigen presentation to T cells and be complementary to the action of ICI. The SOLTI-1503 PROMETEO trial evaluates the effect of T-VEC plus atezolizumab in HER2-negative patients with residual disease on MRI and in a core biopsy after completing standard NAC (50). The combination of neoadjuvant aPD-1, aCTLA-4, and Talimogene Laherparepvec (T-VEC) is currently evaluated in patients with localized TNBC or ER⁺HER2⁻ breast cancer (NCT04185311).

For effective treatment with ICI, it is necessary that tumor-reactive TILs are properly recruited and (re-)activated. Potential hurdles in the efficient activation of TILs through ICI are: potential exhaustion and dysfunction of TILs that limit their capacity to control tumors (51); a (biochemical) barrier that hinders infiltration; or a lack of tumor-reactive TILs or a complete lack of TILs. Several clinical studies aim to support ICI efficacy through an additional therapeutic agent. Combining ICI with IL2 may be one of the future therapies to reinvigorate these antitumor TILs, through enhancing the avidity of PD1⁺CD8⁺ TILs (52), and improved activation and effector function of both CD8⁺ and CD4⁺ TILs (53). Currently, a large phase I study evaluates the safety and antitumor activity of a-PD1 linked to a variant form of IL2 (IL2v) in patients with advanced solid tumors (NCT04303858). In line with this, a phase II neoadjuvant clinical trial on IRX-2, a cell-derived biological with multiple active cytokine components, in combination with aPD-1 and chemotherapy in TNBC is exploring the role of cytokine-based or augmented therapy (NCT04373031). The I-SPY2 trial has reported relevant activity from the neoadjuvant combination of chemotherapy, pembrolizumab, and a TLR-9 agonist in stage II/III breast cancer, leading to an increase in pCR and RCB I outcomes at surgery (54). However, the addition of the TLR-9 agonist did not meet the prespecified threshold for graduation in I-SPY2. As an alternative to reinvigorating dysfunctional TILs, other studies aim to remove factors or cells that may impair optimal TILs function. Upon selection of basal-like TNBCs only in the GeparNuevo, it was shown that immune-rich TNBC with residual disease was dominated by myeloid cells (24). One ongoing phase II clinical trial tests whether the combination of cabiralizumab (inhibitor of CSF-1R) and nivolumab with neoadjuvant chemotherapy will decrease tumor-associated macrophages and increase TILs compared with neoadjuvant chemotherapy plus nivolumab in patients with eTNBC (NCT04331067). Last, T-cell engaging bispecific antibodies have been developed to redirect cytotoxic T cells to predefined tumor antigens, primarily for MHC-independent cancer cell elimination. Simultaneous binding of these bispecific T-cell engaging antibodies to CD3 and the tumor-associated antigen results in the formation of a lytic immune synapse, targeting the tumor cell (55).

The role of adjuvant ICI is now being explored in several clinical trials, as women with residual disease have a higher risk of recurrence than women with other subtypes of breast cancer. IMpassion-030 evaluates the adjuvant addition of atezolizumab versus placebo to adjuvant chemotherapy. The GeparDouze trial (56) investigates whether or not 6 months of adjuvant atezoliumab after neoadjuvant chemotherapy plus ICI affects EFS in patients with eTNBC. In contrast to the women with residual disease that need extra therapy to improve outcomes, women with a pCR after neoadjuvant therapy may not need any additional adjuvant therapy. In the KEYNOTE-522, patients with pCR did well in both arms, with a 3-year EFS of 92.5% with chemotherapy alone and 94.4% with chemotherapy plus pembrolizumab. These findings do call into question the value of the adjuvant phase in those who achieved a pathologic complete response. Whether adjuvant ICI is needed in patients with a pCR will be tested in the upcoming OptimICE-pCR study. These studies will settle the debate regarding the role of the adjuvant ICI component. It will also be critical to evaluate and understand whether or not adjuvant ICI will improve overall outcomes if patients showed a very poor response to neoadjuvant chemotherapy (without ICI) to begin with. Potentially, these patients have a poor endogenous antitumor immune response and may not benefit from additional ICI. The A-BRAVE trial (NCT02926196) and SWOG S1418 (NCT02954874) are two ongoing clinical trials investigating the effect of adjuvant ICI after a non-pCR to chemotherapy.

In addition to the new neoadjuvant combinations, adjuvant combination therapy for patients with residual disease after neoadjuvant treatment for eTNBC is rapidly evolving. Adjuvant olaparib for patients with a germline BRCA mutation not achieving a pCR after neoadjuvant chemotherapy showed significant improvement in DFS (12). The synergy between PARP inhibition and ICI is hypothesized based on increased mutation burden, neoantigen expression, and immunogenic reprogramming (57). Therefore, the combination of adjuvant PARP inhibition with ICI will be a logical next SOC option, especially for germline BRCA mutation carriers. Regardless of BRCA status, the addition of adjuvant capecitabine (11) for patients not achieving a pCR after NAC is of clinical significance. An ongoing phase II trial is currently evaluating the clinical benefit of a combined treatment with radiotherapy and nivolumab plus ipilimumab versus radiotherapy plus capecitabine in eTNBC patients who have residual disease after NAC (NCT03818685). Yet, the combination of adjuvant capecitabine and ICI is not tested in this trial. Importantly, both adjuvant capecitabine or olaparib have not formally been combined with adjuvant ICI in clinical trials for eTNBC. This imposes several risks, such as (unnecessary) toxicity for patients, increasing costs and unclear clinical benefit of the combination of these agents. Nevertheless, safety data from these combinatory therapies are available from studies in patients with metastatic TNBC (mTNBC; refs. 58–60). As suggested by Tarantino and colleagues (61), “pragmatism is warranted” and adjuvant treatment for patients with residual disease may be tailored according to residual cancer burden and germline BRCA status. The addition of adjuvant aPD(L)1 to capecitabine or olaparib could become SOC for patients with eTNBC.

Although ADC data in the adjuvant setting have not been reported yet, these agents seem to be very potent as monotherapy and potentially in combination with ICI as well. Two promising therapeutics include SG and trastuzumab deruxtecan. In the adjuvant setting, the SASCIA (GBG 102; ref. 62) evaluates the use of SG in primary HER2-negative breast cancer, including TNBC, with high relapse risk after standard neoadjuvant treatment. Additionally, in the Destiny-BREAST-04, an impressive survival benefit was observed in patients with metastatic HER2-low breast cancer (IHC 1+, 2+) treated with TDxd over physician's choice chemotherapy, independent of ER expression (45). Future combinations of ICI and ADC, especially in the adjuvant phase after a non-pCR, are potentially interesting, but evidence of synergy has still to be proven. As there are endless possibilities of treatment combinations with novel and regular agents, it is important to rationally design clinical trials, ideally with strong preclinical data supporting the trial's hypothesis.

The current ICI indication for TNBC is still a one-size-fits-all approach, foregoing a tumor- and response-guided approach. First, we need to identify which patients will benefit from the addition of immunotherapy. Second, we need to identify patients who would benefit from immunotherapy whilst optimizing standard (chemo)therapy. Third, we need to identify which patients are not responding to current approaches with ICI and/or chemotherapy. Lastly, irrespective of immunotherapy, we need to acknowledge that a subgroup of TNBC patients exists that even in the absence of any chemotherapy have an excellent survival.

First, we need to identify which patients will benefit from ICI added to chemotherapy. Ideally, this can be predicted from baseline characteristics from the tumor or the blood (Fig. 1). sTILs are an easy and standardized readout for the presence of lymphoid immune cells in the tumor microenvironment of breast cancer (63). Patients with TNBC enriched for sTILs have a higher chance of responding to (neo)adjuvant chemotherapy, and most patients have an excellent prognosis even without the addition of ICI (5, 64). Still, these patients are also likely to respond to ICI as they may have endogenous immune reactivity due to the higher level of sTILs that can be exploited by both chemotherapy and ICI. In the NeoPACT trial, sTILs or immune signatures were associated with high pCR rates approaching or exceeding 80%, yet it is uncertain what the contribution of ICI or chemotherapy is. In the GeparNuevo trial, high sTILs are associated with higher pCR rates in both the placebo and ICI arm, indicating again that it is difficult to study the effect of ICI and chemotherapy separately (16). Nevertheless, an early increase in intratumoral TILs after 2 weeks of ICI was associated with increased pCR rates to ICI and not to chemotherapy alone (16). Expected sTIL data from larger clinical trials, such as KEYNOTE-522 and IMpassion-031, will be crucial in optimizing personalized medicine and patient-centered therapy for future TNBC patients. In GeparNuevo, PD-L1 positivity on immune cells (clone SP263) was only predictive of outcome in the placebo arm, while in IMpassion-031 (clone SP142) and KEYNOTE-522 (clone 22C3), patients with PD-L1–positive tumors had numerically higher pCR rates in both the experimental arms and placebo arms. PD-L1 antibody clones show different staining patterns, and therefore different percentages of tumors are considered positive, making the clinical utility of PD-L1 as a biomarker difficult (65, 66). A high tumor mutational burden (TMB) has been associated with increased pCR rates in all included patients in the GeparNuevo, but this effect seemed to be driven by its predictive value in the placebo arm (67). A recent analysis of these data demonstrated that low TMB was actually associated with improved EFS in the durvalumab arm, suggesting that chemotherapy alone might suffice for patients with eTNBC and high TMB (68). An exploratory analysis of the I-SPY2 trial revealed that clearance of circulating tumor (ct)DNA highly correlated with response to neoadjuvant chemotherapy and survival with or without pembrolizumab (69). These data suggest that sTILs, PD-L1 expression, and TMB are probably not able to accurately distinguish patients who might benefit from ICI from those who already benefit from chemotherapy and that further research on the use of these biomarkers is needed.

Immune-related gene-expression signatures have so far not been able to clearly distinguish response to ICI from response to chemotherapy in eTNBC. In the I-SPY2 trial, the STAT1, mitotic and T-cell/macrophage signatures were associated with pCR after durvalumab/olaparib plus chemotherapy, but the test for interaction with treatment was not significant (70). In GeparNuevo, an IFN-gamma signature and other immune-related signatures were associated with increased pCR in both the experimental and control arms (67, 71); however, recent analysis demonstrated that expression of these signatures was associated with improved EFS only in the durvalumab arm (72), suggesting long-term benefit of ICI in patients with preexisting immune infiltration. As these gene sets only capture part of the complex tumor–immune interactions in breast cancer, more specific gene signatures, for example, capturing the tumor-reactive T-cell pool, might be clinically useful.

Specific assessment of T-cell subsets, T-cell functionality, or the spatial distribution of T cells, might be key to identify patients who benefit from ICI. In a window-of-opportunity study in early-stage breast cancer, it was shown that pretreatment “experienced” CD4⁺ and CD8⁺ T cells (characterized by expression of immune checkpoints and cytotoxicity markers) were associated with clonotypic T-cell expansion upon anti–PD-1, which can be indicative of a clinical response (73). Comprehensive analyses by imaging mass cytometry of the pretreatment samples from the NeoTRIP trial demonstrated that high connectivity between epithelial cells and certain T-cell subsets, such as exhausted T cells and cytotoxic T cells, was predictive of pCR in the experimental arm with atezolizumab (74). In summary, these data on spatial distribution indicate that exhausted and/or cytotoxic T cells near tumor cells might be associated with increased response to neoadjuvant ICI and less likely with chemotherapy or prognosis. Further translational studies, particularly IMpassion-031 and KEYNOTE-522, are essential to validate these biomarkers.

In the neoadjuvant phase, the addition of ICI to standard chemotherapy inspires the question: Can we reduce the current chemotherapy scheme for some patients? Especially the toxicity of chemotherapy, such as fatigue, cognitive dysfunction, neuropathy, and cardiovascular toxicity, implores us to determine what is actually necessary. Currently, we do not know which toxicity, either chemotherapy or ICI related, affects the quality of life the most in the long term, yet it may be crucial information for the patient to opt for optimizing chemotherapy and/or ICI. Preventing overtreatment of eTNBC patients should be considered as a priority in trial design aiming for optimizing therapy in the coming years (Fig. 1). Omitting anthracyclines in a neoadjuvant regimen of taxane/carboplatin plus pembrolizumab seems to be a viable option with a promising pCR rate of 58% in the single-arm NeoPACT trial (and a 98% 2-year EFS rate in those with a pCR; ref. 19) and 48.6% in the randomized NeoTRIP trial (25). Because a high level of sTILs is strongly associated with pCR after neoadjuvant chemotherapy (5) and excellent overall survival without chemotherapy (8, 64), it should be considered to optimize chemotherapy for these patients, which could greatly reduce morbidity without increasing mortality. Additionally, selecting the chemotherapy backbone alongside ICI that can induce optimal immunogenicity and therefore synergy between the two treatments, as explored in the mTNBC TONIC trial (26), may guide treatment optimization in the neoadjuvant setting. Using dynamic biomarkers, one can accomplish the rapid expansion of treatment when, e.g., ctDNA can still be detected during treatment. Currently, there is a lack of accurate imaging tools to monitor early response to ICI. Response to ICI on MRI can be underestimated and therefore is less reliable. Fortunately, tumor uptake and biodistribution of the ICI might play a role in response evaluation. New molecular imaging tracers allow for visualization with positron emission tomography of tumor and immune cell characteristics, which might guide treatment decision-making regarding ICI and chemotherapy in the future (75, 76).

Additionally, reducing the chemotherapy burden can be accomplished in the adjuvant setting. With the introduction of ICI in the adjuvant phase for all TNBC patients, the question arises if adjuvant ICI is actually warranted for patients with a pCR and even if this is rational in the case of a non-pCR. A pCR in TNBC patients is a proven factor for a good prognosis (1) and no other systemic treatment is currently required. In the NeoPACT trial, neoadjuvant aPD-1 plus chemotherapy yielded a pCR of 60% and 2-year EFS of 88% in the absence of adjuvant pembrolizumab. Moreover, whenever an effective immune response is elicited during the neoadjuvant ICI treatment, leading to a pCR, the memory role of the immune system should pertain to this for a long-term effect. Repeating the same ICI treatment, after a good clearance of the tumor and therefore removal of the primary source of neoantigens (post-surgery), does not seem in line with the rationale of immunotherapy.

When a non-pCR is reached, it follows a rationale to expand current therapy. However, it might be debated if repeating the same ICI therapeutics that evidently failed to elicit a pCR in the first place, in the absence now of the bulk tumor and neoantigen source, will be effective. The observed benefit of anti–PD-1 in the RCB II patients in the KEYNOTE-522 could be either due to the adjuvant repeat of ICI or because a more favorable tumor environment and immune response was elicited during the neoadjuvant phase. It is also plausible that the RCB II category contains a very heterogeneous and rather large collection of tumors, thereby allowing for more biological differences and therefore responses. Combinations with other adjuvant therapeutics or therapeutics can kick-start an immune response from a new angle, as proposed in the previous paragraph.

Lastly, irrespective of ICI, we should acknowledge that we may be overtreating a subgroup of TNBC patients who do not respond to ICI or may not need any systemic therapy (Fig. 1). Some patients may not be sensitive to any of the current ICI therapies or combination therapies. Finding predictive biomarkers that can identify patients who will only suffer from adverse events without a tumor response will be crucial to prevent the overtreatment of these patients. Similar to the previous paragraph, tumor-intrinsic biomarkers, as well as sTILs and ctDNA, should be monitored closely at baseline and during therapy to fully understand their potential in detecting a nonresponse. Additionally, two different prognostic studies confirmed that a group of high TIL level TNBC patients exists that have an excellent survival, regardless of therapy (8, 64, 77). TNBC patients with a very high TILs level (≥75%) had a 15-year cumulative incidence of a distant metastasis or death of only 2.1% [95% confidence interval (CI), 0–5.0; ref. 8] and should potentially not be subjected to any therapy expansion but rather (complete) reduction of systemic treatment. Their high level of TILs could potentially lead to a better response to ICI and better response rates in the current clinical trials, yet their overall good prognosis warrants exploration if chemotherapy and ICI can be omitted for these patients.

ICI has recently found its way into regular practice for eTNBC. With improved pathologic complete response rates and improved survival upon ICI, a new era awaits where we exploit the antitumor immune response for this historically difficult-to-treat breast cancer. This era will bring its own challenges and future studies will/should focus on adequate response monitoring for ICI responses, optimizing and reducing chemotherapy burden for selected patients, expansion of adjuvant treatment for patients with a non-pCR, and development of novel immunomodulatory approaches. Making use of baseline or dynamic response evaluation and biomarkers may improve clinical decision-making on tailoring systemic (immune)therapy for TNBC in the future.

M. Kok reports funding to the institute from BMS, Roche, and AstraZeneca/MedImmune and an advisory role for BMS, Roche, MSD, and Daiichi Sankyo outside the submitted work. No disclosures were reported by the other authors.

M. Kok received funding from NWO-Vidi (2020/2021; VIDI-09150172010043) and the Dutch Cancer Society (KWF; 2020-1/12968).

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.