Consensus guidelines for the management of HR-positive HER2/neu negative early breast cancer in India, SAARC region and other LMIC by DELPHI survey method

Although the survey did not include any questions on diagnostic workup, the panelists discussed assessments to be made based on various parameters for each patient during the disease diagnosis. Similarly, opinions on surgery and radiotherapy were consolidated. The steps to be followed for each patient vary on case-to-case basis and the recommendations framed in this survey are in line with guidelines provided by international committees [18]. The diagnostic workup should follow the recommendations mentioned in Table 3. Primary tumor and regional lymph node assessments are key to optimizing therapy in this potentially curative group of EBC. Not all the features are necessary for each patient. Systemic staging of asymptomatic patients is not warranted as routine practice [19]. Further tests are required only when clinically indicated. The bone scan is to be performed in patients with pain in the bone and with elevated alkaline phosphatase if clinically the disease is of stage II [20‐22]. Bone mineral density test is recommended in post-menopausal women who would be treated with aromatase inhibitors for more than 5 years, to avoid the risk of fractures in these women. Abdominal with or without pelvic CT or MRI to be done in cases of abnormal LFT, abnormal physical examination of abdomen/pelvis. Hereditary cancer assessment requires appropriate utilization. The panel experts recommended genetic counselling in women diagnosed with high-risk disease.

The expert committee identified the following factors that are important to assess the risk of recurrence in patients with HR + ve, HER2/neu–ve EBC (Table 1). All the committee members considered nodal status as the most important prognostic factor that also predicts chemotherapy benefit. A total of 96% also considered tumor size to be correlating with the risk of recurrence. For the remaining five features (age, histopathological grade, ER expression, Ki67 levels and gene expression profiling) 88% of experts considered them crucial for predicting the risk of recurrence. The experts opined that tumor size, tumor grade, age of the patient at the time of diagnosis, levels of ER, Ki-67 were more or less of similar importance and correlated with chemotherapy benefit (at least 72% of panelists agreed). Higher clinical risk like big tumors, higher nodal involvement, and moderate and high-grade disease was associated with worst survival and higher recurrence rates [23, 24]. Data from clinical trials have shown that women under 50 years derive a survival benefit of 7–11% with chemotherapy while the benefit of 2–3% in patients aged between 50 and 69 [25]. Besides these factors, the role of proliferation markers, ER and Ki-67 in making systemic adjuvant therapy decisions was discussed. More than 80% of panelists voted for the prognostic role of ER and Ki-67. The threshold for ER positivity has undergone revision from 10 to 1%, with data suggesting a limited endocrine therapy benefit in patients with ER lower than 10% [26]. Nonetheless, panelist agreed to 1% of ER as being positive honouring ASCO and CAP recommendations [27]. However, PR status failed to achieve a positive consensus threshold in both rounds 1 and 2 and hence was excluded from the table. In view of the subjective nature of Ki-67 staining and grading, 80% of panelists expressed their opinion that expression levels of Ki-67 of 14% and above are considered high-risk. The latest recommendations from the International Ki-67 in Breast Cancer working group opined that Ki-67 above 30% could be considered high-risk and for patients with Ki-67 between > 5-<30% advice from a prognostic test should be considered for deciding on chemotherapy use [28]. The panelists discussed the importance of gene expression profiles for decision-making on the use of chemotherapy at length. 88% of the panelists voted for the use of prognostic tests that assess the risk of recurrence based on the expression of genes that provide significant information on cancer progression which clinical parameters and proliferation indices might miss out [29]. Hence the consensus guidelines statement is that all seven features are important for assessing the aggressiveness of the disease thereby predicting the risk of recurrence and to be considered for making decisions on whether to use chemotherapy or not (Table 2).

The expert committee evaluated three online predictive tools – NPI, IHC4 and PREDICT. The opinion on their utility was variable (Table 1). NPI uses a simple equation that estimates overall survival based on clinical parameters while PREDICT along with clinical parameters uses information on age, menopausal status, ER and Ki-67 for making survival estimates post-surgery for various adjuvant treatment regimens [30, 31]. Contrary to these two tools, IHC4 is purely based on information derived from immunohistochemistry of ER, PR, Ki-67 and HER2 without considering the clinical parameters [32]. Even after extensive discussion and re-review of updated literature, there was no confidence in the value of such predictive tools. Although these online tools provide reliable prognostic information in some cohorts, they have been shown to overestimate or underestimate survival in patients of certain age groups. NPI is sub-optimal in predicting prognosis in patients < 40 years and underestimated overall survival in patients aged between 55 and 60 years [33]. In a study on a cohort of 600 patients, aged below 40 years PREDICT overestimated chemotherapy benefits and 10-year mortality by 8% [34, 35]. Moreover, NPI and IHC4 have ambiguous intermediate-risk zones, failing to provide a definite treatment option to these patients. The panelists highly opined cancer recurrence and progression are driven by key biological markers, which manifest in upstaging of tumor anatomical features by transforming clinical low risk into high risk. AJCC revised the staging definition from anatomic staging to a prognostic staging system with the incorporation of biomarkers in its 8th edition [36]. As a result, a recent study conducted on 4729 patients with T1-T2N1mi disease with 5-year follow-up found that approximately 84.4% of patients were downstaged and 3.7% of patients were upstaged and reported that 8th AJCC system predicted better breast cancer-specific survival compared to 7th AJCC staging system [37]. Hence the consensus guidelines statement is that these online predictive tools that primarily use clinical parameters to predict survival benefits, are not to be used in patient decision-making (Table 2).

The expert committee spent substantial time discussing the role of multi-marker prognostic tests. Five tests were considered for the discussion in this survey (Table 1). A total of 84% of experts were in favour of CanAssist Breast and 80% were also in the favour of Oncotype DX. The usage of the other three tests ranged between 20% and 28% of the experts. Oncotype DX and MammaPrint are the first-generation prognostic signatures. Oncotype DX is a 21 gene signature that has been developed in the NSABP-14 cohort and validated in NSABP-20, TransATAC and SWOG8814 cohorts. Other than these, Oncotype DX has been validated in large clinical trials in a prospective manner in patients with node-negative (TAILORx trial) and node-positive patients (RxPONDER). MammaPrint is a 70 gene signature validated in prospective trials, RASTER and MINDACT. Prosigna is a 50-gene signature initially developed to distinguish luminal subtypes. This test is validated only in post-menopausal women with different risk zones for node-negative (3 zones) and node-positive patients (2 zones). EndoPredict is a 12-gene signature that later along with clinical parameters (node and tumor size) was called EPClin. This test has been validated retrospectively on ABCSG6 and 8 cohorts [29]. While all the above mentioned tools use logistic regression, CanAssist Breast is an IHC based tool that uses an Artificial Intelligence-based Machine (SVM) algorithm for the prediction of recurrence risk with inputs from immunohistochemistry information of 5 critical biomarkers of cancer progression and recurrence pathways along with 3 clinical parameters [38]. CanAssist Breast has been validated in multiple cohorts from Southeast Asia, the USA and Europe [39‐41] and in a DUTCH sub-cohort of TEAM trial- a prospective randomized trial [42]. The consensus guidelines state that CanAssist Breast and Oncotype DX are the recommended multi-marker prognostic tests that are the preferred choice (Table 2).

Other more recent prognostic tools which are not considered in this survey are DigiStain and a test developed in Korea named GenesWell BCT [43, 44]. Both these tests are relatively new in SAARC countries; DigiStain is available for patients in select regions and the Korean test is undergoing additional validation studies to be available for patients, we can perhaps include them in the survey in the next updating of the guidelines.

The appropriate clinical utility of multi-marker prognostic tests was also discussed. A total of 92% of the experts agreed that their use is based on patient affordability and on inclusion in international treatment guidelines (Table 1). An identical number (92%) also agreed that Western tests have not been validated in Indian patients; also 88% said that the TAILORx trial that assessed Oncotype DX performance in node-negative patients included a few Indian/Asian patients and hence is not applicable to this ethnic/geographical group [45]. While 88% used multi-marker prognostic tests routinely and 84% said they recommended use when facing a clinical dilemma in treatment decision- making. The consensus guidelines statement is that multi-marker prognostic tests should be used routinely based on patient affordability and especially when facing a clinical dilemma. Even if Western tests are included in international guidelines they should not be used blindly since they have not been validated in Asian and Indian patients (Table 2).

Aligning with the data published from multiple cohorts, 76% of the panelists agreed that differences exist in breast cancer between Asian and Western women and therefore would require different treatment approaches [46]. Asian women are diagnosed more at a premenopausal age, with larger luminal B tumors, with node-positive disease, and have active tumor micro-environments with frequent TP53 mutations vs. Caucasian women warranting aggressive treatment strategies [47‐50]. The consensus was that Asian (including Indian) younger women have the more aggressive disease (84%); high expression of proliferative genes (80%); and higher involvement of genes involving endocrine resistance (80%) (Table 1). In a study involving a large SEER database of 86,030 patients who underwent Oncotype DX testing, it was found that a large number of Black women were likely to have RS greater than 25 (high-risk requiring chemotherapy) compared to White women with lower accuracy of Oncotype DX in identifying low-risk patients (RS 0–25) [51]. This data along with post-hoc analysis of TAILORx showed differential performance of Oncotype DX in Black vs. White women, a total of 88% agreed that Black women had higher mortality compared to White women of the same type, grade and stage of breast cancer within the same RS score. The consensus guidelines statement is that Asian/Indian patients have a biologically different disease which is more aggressive in younger patients and can have higher expression of poor prognostic genes compared to Western women (Table 2).

Regarding CanAssist Breast as a prognostic and predictive test, 84% confirmed that it predicted risk of recurrence based on tumor biology; 80% said it is affordable in LMIC; and 76% stated that it predicts risk of recurrence across diverse ethnic backgrounds and geographies (Table 1). CanAssist Breast is the only test that has been extensively validated on Southeast breast cancer patients [36‐38]. Along with this data, CanAssist Breast has validation data on Caucasian patients, who are different from its development cohort (Indian). This data clearly demonstrated its unparalleled performance in the cohorts from Europe and USA [41]. Despite the differences in the disease (breast cancer) between Asian and Caucasian women, CanAssist Breast’s prognostication was similar across these diverse cohorts unlike Oncotype DX, MammaPrint. Post-hoc analysis of the TAILORx trial showed altered performance of Oncotype DX with a higher hazard ratio in Black women vs. White women and non-Hispanic women vs. Hispanic women for the same RS category [45]. Likewise, MammaPrint showed lower low-risk proportions in Asian patients compared to European breast cancer patients [52]. Along with this outstanding performance across the various cohorts, CanAssist Breast showed greater than 83% concordance in the low-risk category with Oncotype DX and MammaPrint, the greatest agreement shown between two prognostic tests ever [41, 53]. Optima prelim trial that assessed the agreement between Oncotype DX, MammaPrint, Prosigna, IHC4 and IHC4 Aqua reported a disagreement between these tests in 60.6% of tumors [54]. The accuracy of the test and its cost are the major determining factors in the choice of a prognostic test. With the five times higher price of Oncotype DX and MammaPrint [55, 56] than that of CanAssist Breast [40], the use of these Western multi-gene tests for many patients from India, SAARC countries would be a far-fetched option. The consensus guidelines statement is that CanAssist Breast is a prognostic test that predicts the risk of recurrence; is applicable across ethnic backgrounds and geographies and is affordable in LMIC (Table 2).

The use of hormonal therapy for more than 5 years in postmenopausal women reduced the risk of recurrence as well as the risk of contralateral breast cancer was the opinion of 92% of panelists (Table 1). These opinions were based on the results of a placebo-controlled clinical trial and ATLAS trial [57, 58]. In line with the results from aTTOM trial [59] where BCI (Breast Cancer Index) high-risk patients derived significant benefit from extended hormonal therapy, 80% of panelists voted for the opinion that recurrence after 5 years of adjuvant endocrine therapy is seen in patients who have been identified as having high risk based on multi-marker prognostic testing. Another factor worth considering for treating post-menopausal women with extended hormonal therapy is bone health. The results of bone mineral density test will be a guiding factor for prescribing hormonal therapy beyond 5 years. All members of the expert committee agreed that hormonal therapy should be used for male patients (Table 1) and 96% stated the preferred choice of treatment is tamoxifen. The consensus guidelines statement is that hormonal therapy with tamoxifen should be used in male patients; use of hormonal therapy for more than 5 years reduces the risk of recurrence as well as contralateral breast cancer; recurrence after 5 years of endocrine therapy occurs in patients who have been identified as having high risk by multi-marker prognostic testing (Table 2).

Although there were no questions asked in the survey regarding luminal sub-types, in the final in-person round the panelists discussed the adjuvant systemic treatment strategies for luminal-like patients (Table 4). The panelists concluded that luminal-A like patients have a good prognosis with endocrine therapy alone. Few patients are identified as ‘high-risk’ by the multi-marker prognostic tests and patients with high tumor burden with large tumors or high nodal involvement or poorly differentiated tumors derive benefit from chemotherapy. Patients whose tumors are Luminal-A like are generally suitable for endocrine therapy alone. In case they have high-risk features (on biomarker testing or high tumor burden), chemotherapy should be added. St Gallen’s expert panel also recommends chemotherapy for luminal-A patients with large tumor volume [60]. For patients who have Luminal-B like tumors and are HER2/neu negative, both chemotherapy and endocrine therapy are to be used. If biomarker testing indicates a low risk for cancer recurrence, chemotherapy can be avoided in such patients. Oncotype DX and CanAssist Breast are known to identify high-risk patients from luminal-A like patients and low-risk patients from luminal-B like patients [40, 61].