Real-World Outcomes with the KEYNOTE-522 Regimen in Early-Stage Triple-Negative Breast Cancer.
Oct 22, 2024DOI: 10.1245/s10434-024-16390-7Journal ArticleASO
Written by: Casey ConnorsStephanie A ValenteAyat ElSherifPaula EscobarAnna ChichuraLauren KopickyErin RoeschJulie RitnerPatrick McIntireYueqi WuChao TuJulie E Lang
This study was approved by the Cleveland Clinic Institutional Review Board (IRB). We retrospectively reviewed our prospectively maintained tumor registry at a single academic institution. All female patients diagnosed with stage II–III TNBC between January 2019 and December 2022 who underwent NAT and subsequent surgical intervention were included. Staging was based on the American Joint Committee on Cancer staging criteria (eighth edition).17We recorded clinical parameters including age, race, body mass index (BMI), Eastern Oncology Group (ECOG) performance status, genetic mutation, tumor grade, clinical and pathologic tumor stage, clinical and pathologic nodal stage, candidacy for BCT pre- and post-NAT, magnetic resonance imaging (MRI) pre- and post-NAT, and occurrence of irAE classified as grade 2 or higher as well as the specific type of irAE. All information was extracted from electronic medical records.As there is no standardized definition of qualification for BCT, for this study, patients were deemed eligible for BCT if on clinical imaging, their solitary tumor measured < 4 cm, or the total measurement of multifocal disease was < 4 cm and there was no evidence of multicentric disease or inflammatory breast cancer. Breast MRI was obtained at the surgeon’s discretion. The decision for a particular NAT regimen was at the discretion of the medical oncologist based on the most updated National Comprehensive Cancer Network (NCCN) guidelines. The KEYNOTE-522 regimen became the standard of care at our institution in July 2021 after the US Food and Drug Administration (FDA) approved pembrolizumab for use in the neoadjuvant setting for patients with stage II/III TNBC. In patients who received NAT before the approval of the KEYNOTE-522 regimen and for those who were not candidates for immunotherapy, doxorubicin, and cyclophosphamide followed by paclitaxel was the most commonly used chemotherapy regimen. Patients were deemed ineligible for immunotherapy if they were pregnant during treatment, actively receiving immunosuppression or had a known history of select autoimmune disorders. For patients not medically fit to receive doxorubicin, a combination of docetaxel and cyclophosphamide chemotherapy was administered.Surgery was scheduled for 4–6 weeks after the completion of NAT. Axillary management was at the surgeon’s discretion. Ultrasound of the axilla was strongly encouraged at our institution, but not mandatory. All patients with suspicious appearing axillary lymph nodes underwent pre-NAT axillary staging with ultrasound-guided core needle biopsy with placement of a post-biopsy clip. Sentinel lymph node biopsy (SLNB) was performed for patients with clinically negative axillae and at the surgeon’s discretion for patients with biopsy-proven metastatic nodal involvement with clinical improvement after NAT. When targeted axillary dissection was performed, successful retrieval of all clipped nodes was confirmed with radiologic and pathologic identification. Axillary lymph node dissection (ALND) was performed on all patients with a diagnosis of inflammatory breast cancer and those with residual nodal disease after NAT. Two patients who underwent axillary surgery only for occult primary breast cancer in the KEYNOTE-522 group were excluded from analysis involving surgical procedure and in-breast response to NAT.
The primary outcome was BCT rate. Secondary outcomes included candidacy for BCT, pCR rate [defined as no residual invasive disease in the breast and lymph nodes (ypT0/is, ypN0/Ni+)], identification of clinical characteristics to help predict pathologic response to NAT, identification of factors that influence surgery type, ability of MRI to predict pCR, mean tumor reduction (in the event of multiple tumors the largest tumor dimension was reported), and irAEs.
Event-free survival was defined as the time from diagnosis to the date of local or distant recurrence or death due to any cause, whichever occurred first.
Data from 240 patients were collected. The chi-squared test or Fisher’s exact test were performed to compare the associations with categorical variables between groups. The Kruskal–Wallis test was used to compare the mean change in tumor size between groups before and after NAT. A multivariate logistic regression model was used to identify significant factors associated with surgical procedure. Stepwise logistic regression was used to select fewer predictor variables for building the best performing logistic regression model. The Kaplan–Meier method was used to estimate EFS and OS with differences between cohorts compared with the log-rank test. Analyses were performed using GraphPad Prism version 10 and R Studio software. A p value < 0.05 was considered statistically significant.
RESULTS
A total of 240 female patients diagnosed with stage II–III TNBC who underwent NAT followed by surgery at our institution were identified. A total of 86 patients received the KEYNOTE-522 regimen, whereas 154 patients underwent NAT with standard of care chemotherapy (“control group”). In the KEYNOTE-522 cohort, 78% completed the intended regimen, versus 86% for the control. Patient, tumor, and treatment characteristics for the KEYNOTE-522 and control groups are shown in Table 1. Median patient age was 56 years old, 66% (158/240) had T2 tumors, 83% (200/240) were grade 3 via Nottingham grading score, 46% (110/240) had clinically involved lymph nodes at diagnosis and 79% (190/240) of patients presented with a self-detected mass on presentation.
The frequency of pCR (ypT0/is, N0/Ni+) was significantly higher in the KEYNOTE-522 compared to the control group, 59.3%, and 33.1%, respectively [odds ratio (OR) = 0.34; 95% confidence interval (CI), 0.20 to 0.58; p = 0.001]. Axillary nodal involvement at the time of biopsy was seen in 40 patients (46.5%) in the KEYNOTE and 70 patients (45.5%) in the control group with subsequent SLNB performed in 73 (84.9%) and 118 (76.6%) patients, respectively. Patients who received the KEYNOTE-522 regimen were more likely to achieve ypN0 status (80.2% versus 68.8%; p = 0.06). Of the patients with a biopsy-proven nodal metastasis at diagnosis, 67.5% of those who received KEYNOTE-522 achieved a complete nodal response (ypN0/Ni+) compared with 44.3% in the control (p = 0.02). Therefore, the rate of ALND was significantly lower after KEYNOTE compared with control (39.6% versus 25.6%, p = 0.03).
Patient candidacy for BCT based on tumor size at initial presentation was 46.4% (39/84) in the KEYNOTE-522 and 39.6% (61/154) in the control group which increased to 75.0% (63/84) and 70.8% (109/154) after NAT, respectively. Despite the significant difference in pCR rate with KEYNOTE-522, there was no significant difference in the BCT rate between the KEYNOTE-522 versus control, 32.1% (27/84) versus 33.1% (51/154), respectively (p= 0.47) (Fig. 1).
Figure 2 displays a Forest plot for multivariable analysis of factors associated with BCT versus mastectomy surgery. Age < 50 years was the only factor significantly associated with mastectomy (OR 0.40; 95% CI 0.20–0.76; p = 0.02).
Reduction in mean tumor size between the two groups was compared using largest tumor measurement on pre-NAT imaging to the largest size on final pathology (Fig. 3). The mean tumor size at diagnosis in the KEYNOTE-522 and control groups were 4.01 cm and 4.19 cm, respectively. There was a 79.8% (3.2 cm [95% CI 2.02–4.27]) decrease in mean tumor size following NAT with the KEYNOTE-522 regimen versus 66.7% reduction (2.80 cm [95% CI 1.60–4.0]) with the control treatment. The difference in the mean change between the two groups was not statistically significant (p = 0.28).
The ability of MRI to predict pCR was evaluated. Of the 158 patients who underwent post-NAT MRI, 69% achieved a complete imaging response (CIR). Of the 69 patients with a CIR, 54 (78.3%) achieved a pCR. For those patients with a CIR who did not achieve a pCR, the largest residual tumor size on final pathology was 0.7 mm (range 0.1–0.7 mm).
We also investigated the association between pCR and histologic subtype. Of 14 patients diagnosed with IDC with lobular features via core needle biopsy, only 1 patient (7.1%) in either group achieved a pCR. On multivariate analysis, invasive ductal carcinoma (IDC) with lobular features was predictive of not achieving pCR (OR 8.65; 95% CI 1.36–17.2; p = 0.02). All four patients with the presence of tumor-infiltrating lymphocytes (TILs) on core needle pathology who underwent NAT with the KEYNOTE-522 regimen achieved a pCR. This feature was not significant in the predictive model given the low number of patients.
At a median follow-up of 2.4 years, there was no significant difference in the rate of event-free survival defined as locoregional recurrence, distant recurrence or death from any cause between the KEYNOTE-522 and control (Fig. 4) (hazard ratio 0.59; 95% CI 0.31–1.14; p = 0.12) The two groups had no significant difference in disease-free survival (88.4% in the KEYNOTE-522 group and 74.0% in the control group), recurrence-free survival, (93.0% and 80.5%), and overall survival (95.3% and 83%).
Genetic testing revealed pathogenic variants in 19% (45/240) of patients including BRCA1, BRCA2, PALB2, CHEK2, ATM, and RAD51D. BRCA1/2 accounted for 62% of all pathogenic mutations. In the subset of 17 patients with a pathogenic BRCA1 variant, there was no significant difference in pCR rate between those who received the KEYNOTE-522 regimen versus control, 80.0% (4/5) and 75% (9/12), respectively (p = 1). Patients harboring a BRCA1 mutation who underwent NAT with the control regimen were significantly more likely to achieve a pCR compared with wild–type BRCA1 carriers (75.0% versus 29.6%) (p = 0.003). There was no difference in polymerase chain reaction (PCR9 rate for patients treated with the KEYNOTE-522 regimen in the subset with a BRCA1 mutation (80% versus 55.6%) (p = 0.39).
The incidence of adverse events (grade 2 or higher) attributable to pembrolizumab was 34.9% in patients who underwent the KEYNOTE-522 regimen. It was noted that 10 (33.3%) patients diagnosed with irAEs experienced multiorgan involvement. A total of 42 irAEs were identified among the 30 patients affected (Fig. 5). Endocrinopathies, including hypo/hyperthyroidism and adrenal insufficiency, accounted for 47.6% of all irAEs. No new cases of diabetes mellitus or hypophysitis occurred due to irAEs. The development of irAEs occurred with relatively equal frequency in the neoadjuvant (22/42, 52%) and adjuvant treatment phase. There was no significant correlation between age, BMI, race, tobacco use or attainment of pCR in the development of irAEs in our patient population.
Baseline levels of thyroid stimulating hormone (TSH) and cortisol were obtained in 76.6% and 36.7% of patients, respectively (of n = 30 with evaluable laboratory studies for irAEs). Additionally, nine patients had no cortisol level drawn before operation since they were treated early in our institutional experience prior to standardizing protocols to incorporate TSH and cortisol levels.
DISCUSSION
We conducted a retrospective review to evaluate the rate of BCT for patients with early-stage TNBC who received NAT with KEYNOTE-522 regimen compared with those who underwent standard of care chemotherapy. We found a statistically significant difference in pCR rate of 59.3% versus 33.1% with KEYNOTE-522 versus control. These findings demonstrate real-world pCR rates comparable with the 64.8% rate (95% CI 59.9–69.5) seen in the KEYNOTE-522 trial,13 suggesting that those results are generalizable.
Despite this higher pCR rate, the hypothesized increase in BCT in patients undergoing NAT with chemoimmunotherapy was not observed, although most patients in the study were BCT candidates. Notably, candidacy for BCT after NAT increased by greater than 50% with both regimens. Patients with multicentric disease were not considered candidates for two-site BCT as our study was prior to the publication of the ACOSOG Z11102 results.18 The reduction in mean tumor size between the two groups, 79.8% reduction in the KEYNOTE-522 cohort and 66.7% reduction in the control group, were also statistically insignificant. The KEYNOTE-522 trial did not report data on mean tumor size reduction.
We performed a multivariate analysis to identify factors associated with surgical procedure type. Our results demonstrate that age < 50 years was predictive of choosing mastectomy. Sun et al.19 reported trends in surgical procedures over multiple decades for young patients with breast cancer, demonstrating an overall BCT rate of 34.6% in 10,312 patients less than 40 years old with early-stage breast cancer. The results were more pronounced in patients found to be estrogen receptor (ER)-negative compared with those diagnosed with ER-positive (ER+) tumors (32.4% versus 67.6%) (p = 0.001).19 This demographic trend of younger patients opting for mastectomy may partially explain our findings, since TNBC often affects a younger population.20 The low BCT rate in our patient cohort was also observed by Fayaz et al.,21 who found a BCT rate of 33% among 359 women diagnosed with TNBC.
Another key finding in this study is the significant reduction in ALND in patients who received the KEYNOTE-522 regimen, from 39.6% to 25.6%. With the high rate of pCR in the axilla, the KEYNOTE-522 regimen has been pivotal in allowing surgical deescalation of the axilla in patients with TNBC. Similar to the KEYNOTE-522 trial, patients who received KEYNOTE-522 at our institution were significantly more likely to obtain both nodal pCR (ypN0/Ni+) (67.5%) and in-breast pCR (ypT0/Tis) (60.7%).
Trends were noted between histologic subtypes and pCR and thus were included in a predictive model. We found that the pathologic subtype of invasive ductal carcinoma (IDC) with lobular features on core needle biopsy was predictive of not achieving pCR. The same difference was not observed in patients with invasive mammary carcinoma with mixed ductal and lobular features. In contrast, all patients with tumor-infiltrating lymphocytes (TILs) identified on CNB who received KEYNOTE-522 achieved a pCR, although this was not statistically significant given the small sample size. This is in line with the publication by Bairi et al.,22 which supports the use of TILs as predictive and prognostic biomarkers in patients receiving immunotherapy. Larger data sets are needed to demonstrate the true impact of histologic subtypes on response to the KEYNOTE-522 regimen.
Our study had a short median follow-up of 2.4 years and thus failed to demonstrate a significant increase in OS or EFS for the KEYNOTE-522 regimen. The EFS in this study (88.4% in the KEYNOTE-522 group and 74.0% in the control) is very similar to the 3.3-year follow-up data from the KEYNOTE-522 trial (84.5% compared with 76.8%).13 Our findings were also similar to the experience with pembrolizumab plus chemotherapy from I-SPY2 (93% EFS at 3 years with 2.8 years median follow-up).10 As the data mature, EFS may become significant over time.
Our findings provide real-world data on pCR rates and surgical preference for patients with early-stage TNBC undergoing NAT with the KEYNOTE-522 regimen at a large academic institution. The KEYNOTE-522 regimen provides superior rates of pCR but risks chronic illnesses associated with irAEs. Clinicians highly value pCR since pCR correlates with improved EFS; however, surgeons should take note of the high rates of irAEs.6,7 Because it was difficult to capture grade 1 irAEs, especially in the early stages of implementing the KEYNOTE-522 regimen at our institution, all of the irAEs in this data were ≥ grade 2. Approximately one-third of patients who received pembrolizumab developed irAEs. At our institution, a laboratory irAE screening protocol for patients undergoing NAT with KEYNOTE-522 was being developed while this cohort was accrued. Prior to standardization of screening tests throughout the Cleveland Clinic enterprise, laboratory data analyzed from patients diagnosed with irAEs demonstrated that baseline TSH and cortisol were not consistently drawn prior to initiation of NAT. Additionally, 30.0% of patients being treated with pembrolizumab in the neoadjuvant setting had no cortisol level drawn prior to surgical intervention, two of whom were diagnosed with adrenal insufficiency in the immediate postoperative period. This data highlights the importance of both surveillance and collaboration of the entire multidisciplinary team, as missed irAEs pose a significant risk in the perioperative period. An institutional order set that incorporates TSH and cortisol prior to surgical intervention to screen for irAEs is advised. Our institution uses guidelines to monitor for endocrinopathies, which have been modified from current NCCN guidelines.23
Numerous risk factors for susceptibility to irAEs from ICIs have been previously reported including chronic smoking, age < 60, high BMI, and achievement of pCR on final pathology.24,25 In our patient population, none of these risk factors were significant in predicting irAEs. Myers et al.26 evaluated the impact of chemoimmunotherapy on surgical outcomes for patients with operable TNBC, identifying irAEs in 43% of the 139 patients undergoing chemoimmunotherapy. This number is higher than the 33% seen in our study as we excluded grade 1 irAEs due to the inability to rule out chemotherapy as the offending agent particularly early in the use of chemoimmunotherapy at our institution.13,26 Endocrinopathies accounted for 48% of all irAEs in our study, which is higher than the 20.4% published in the KEYNOTE-522 trial.13
In the subset of patients with a germline BRCA1 mutation, there was no significant difference in pCR rates between those who received the KEYNOTE-522 and those who received the control regimens. A difference was likely not observed due to the excellent pCR rate of 75% in BRCA1 carriers treated with the control regimen. There are no prior studies comparing pCR rate among BRCA1 carriers undergoing the KEYNOTE-522 regimen and those receiving anthracycline- and taxane-based regimens. A systematic review and meta-analysis by Jia et al.27 evaluated 22 studies that included patients with TNBC and identified 392 patients harboring BRCA1/2 gene mutations. The authors noted a 17% increase in pCR rate over non-BRCA1/2 carriers when platinum-based chemotherapy was utilized.27 Many studies often combine BRCA1/2 variant carriers given the overall low number of patients included.27 In our study, we found that patients with a BRCA1 mutation achieved exceptionally high pCR rates even in the absence of platinum- and immunotherapy-containing NAT regimens, which was not observed in patients with BRCA2 mutations. Should additional studies confirm these findings, consideration should be made for sparing triple-negative BRCA1 patients from immunotherapy.
Our study does have some limitations. As with all retrospective cohort studies, our study was subject to selection bias. Also, the choice of NAT regimen was made at the discretion of the medical oncologist. The subset analysis regarding BRCA1 carriers is undoubtedly a small population but it is hypothesis-generating that this population of TNBC patients might have excellent response rates to standard NAT. Additionally, the use of adjuvant pembrolizumab is not identical to the KEYNOTE-522 trial for non-pCR patients at our institution, which may or may not impact survival outcomes and rates of irAEs. The scope of our study did not focus on the impact of adjuvant capecitabine or olaparib for BRCA carriers on outcomes. Targeted axillary dissection (TAD) was more recently adopted at our institution, therefore the non-KEYNOTE cohort appeared to show a time bias of less use of TAD.
We found that the KEYNOTE-522 regimen, when utilized outside the setting of a clinical trial, provided a comparable significant improvement in pCR rate for patients with stage II/III TNBC undergoing NAT. Despite the improved pCR, there was no difference in rates of BCT or BCT candidacy between those who received NAT with or without immunotherapy. Patients who received NAT with KEYNOTE-522 had a significant reduction in ALND. The addition of pembrolizumab produced irAEs ≥ grade 2 in one-third of patients, with endocrinopathies being the most common. This study is one of the first to report a real-world experience of the KEYNOTE-522 regimen on surgical outcomes. These results provide insight into the surgical management of patients with stage II/III TNBC after the KEYNOTE-522 regimen.
LK
Dr. Lauren Kopicky
Surgery · Breast Surgical OncologyCleveland Cliniclaurenk