Breast cancer is the most commonly diagnosed cancer and the second leading cause of cancer-related deaths in the United States (1). Human epithelial growth factor receptor 2 (HER2) is overexpressed in ~15% of breast cancer cases in the United States (2), and HER2-positive breast cancer is more common in metastatic settings (3). In breast cancer treatment, the development of targeted therapies has improved the efficacy, reducing damage to normal tissues; however, increased drug resistance against targeted agents has led researchers to develop an antibody-drug conjugate comprising cytotoxic agents and monoclonal drugs (4). Trastuzumab emtansine (T-DM1) is the first antibody-drug conjugate to be approved in a solid tumor and consists of trastuzumab, humanized monoclonal antibodies against HER2 and mertansine, a microtubule inhibitor (4). Based on the results of the EMILIA (5), TH3RESA (6) and MARIANNE (7) trials using various patient groups, T-DM1 has been approved for patients with HER2-positive breast cancer (HPBC) that have previously been treated with trastuzumab and taxane. Although T-DM1 can be used for any line of treatment, it is accepted as one of the standard second-line regimens for the treatment of metastatic HPBC (8,9).

Thrombocytopenia is one of the adverse events observed during T-DM1 treatment and is the most common reason for dose reduction and treatment discontinuation (10–13). Although previous reports assert the opposite (14), recent studies have reported that a number of systemic toxicities, including thrombocytopenia, may predict T-DM1 efficacy (15,16). As it has been reported that the incidence of thrombocytopenia increases with prolonged T-DM1 treatment duration (17), the predictive value for survival outcome is debatable. Thus, the present study aimed to investigate whether early thrombocytopenia during T-DM1 treatment could predict survival rates.

The present study enrolled 138 patients and Table I presents the characteristics of the patients and tumors. The median age was 50 years (range, 26–83 years). Of the 138 patients, 58% were hormone receptor-positive, 83% had a HER2 score of 3+ in the IHC evaluation and 50% were post-menopausal. The majority of the patients (77%) presented with visceral metastasis and 41% had de novo metastatic disease. The most common site of metastasis was bone (55%), followed by the lung (50%) and non-regional lymph nodes (45%). The ECOG performance status score was <2 in 90% of the patients. Before T-DM1 administration, all of the patients had received trastuzumab, and 95% had received taxanes. Pertuzumab was administered to only 17% of the patients before T-DM1. The median number of lines of treatment for T-DM1 was 2 (range, 1–8) in metastatic cases.

Tables II and III present the efficacy outcomes and laboratory toxicities of treatment with T-DM1, respectively. The median number of cycles was 9 (range, 2–58), and 33% of the patients received ≥15 cycles of T-DM1. Dose reduction for the subsequent cycle was performed in 12 patients (9%), and half of those were due to thrombocytopenia. At the time of data cut-off, 86% of the patients had experienced TTD events. The ORR and DCR were 50.0 and 69.6%, respectively. The most common all-grade adverse events were increased levels of hepatic enzymes (43 and 38% for AST and ALT, respectively), thrombocytopenia (39%) and anemia (38%). Among grade 3 and 4 toxicities, thrombocytopenia was the most common adverse event (10%). The median number of treatment cycles in which thrombocytopenia first appeared was 3 (range, 1–32) (Fig. 1A), and two-thirds of occurrences of thrombocytopenia (66.7%) were observed in the first four cycles (Fig. 1B).

The median follow-up time was 19.3 months (range, 1–70 months). Based on the Kaplan-Meier analysis, the median TTD was 9.5 months [95% confidence interval (CI), 7.3–11.7] (Fig. 2A), and the median OS was 28.2 months (95% CI, 19.2–37.2) (Fig. 2B). Table IV presents univariate and multivariate Cox regression analyses for factors affecting the TTD of T-DM1. The multivariate analyses revealed that hormone receptor status [hazard ratio (HR), 1.837; 95% CI, 1.249–2.701; P=0.002], ECOG performance status score (HR, 3.269; 95% CI, 1.788–5.976; P<0.001) and thrombocytopenia during treatment (HR, 0.297; 95% CI, 0.198–0.446; P<0.001) were independent factors for TTD. Patients who were hormone receptor-positive and patients with high ECOG performance scores had an increased risk of an TTD event. By contrast, patients who developed thrombocytopenia during T-DM1 treatment had a reduced risk of developing a TTD event.

Fig. 3 presents the TTD survival curves according to thrombocytopenia. Patients with thrombocytopenia during treatment had a longer TTD (P<0.001) (Fig. 3A). As a longer duration of T-DM1 treatment was associated with a high incidence of thrombocytopenia, a further analysis was performed to evaluate whether thrombocytopenia was predictive of a prolonged survival time. A comparison of the survival rates between patients with and without early thrombocytopenia using a log-rank test revealed that patients with early thrombocytopenia had a significantly longer TTD of 17.3 months (95% CI, 11.8–22.8) compared with 7.6 months (95% CI, 5.8–9.4) for patients without early thrombocytopenia (P<0.001) (Fig. 3B).

The present study evaluated the factors affecting survival time during T-DM1 treatment. It was revealed that increased survival time was associated with low ECOG performance scores, negative hormone receptor status and thrombocytopenia. Furthermore, thrombocytopenia that developed in the first four T-DM1 cycles was predictive of a longer TTD in T-DM1 treatment.

Thrombocytopenia has been identified as a characteristic of T-DM1 and as the most common grade ≥3 and dose-limiting adverse event in both clinical trials (10–12) and real-life studies (17,24,25). To the best of our knowledge, only two reports in the literature have studied the predictive value of thrombocytopenia during T-DM1 treatment. Tataroglu et al (16) reported that a multivariate Cox regression analysis of 78 patients demonstrated that patients with thrombocytopenia had longer progression-free survival (PFS) compared with those without thrombocytopenia, consistent with the results of the present study. In the second study, which evaluated 73 patients, Tang et al (15) proposed a toxicity score including thrombocytopenia and hepatitis. It was revealed that an increased toxicity score was associated with an improved response and prolonged PFS. However, Yardley et al (17) observed that the incidence of thrombocytopenia increased with T-DM1 cycles, and thrombocytopenia was more often observed in patients receiving >18 cycles of T-DM1 compared with in those receiving ≤18 cycles.

Considering that the incidence of thrombocytopenia increases with the duration of T-DM1 treatment, it is a matter of debate whether thrombocytopenia is an inevitable adverse event secondary to long-term use or is a predictive marker for improved survival rates. To investigate this issue, the survival rates of patients with thrombocytopenia in the first four T-DM1 cycles were compared with those without thrombocytopenia in the first four cycles after thrombocytopenia was confirmed as a significant independent factor for survival rate using a multivariate analysis. It was revealed that patients with early thrombocytopenia exhibited improved survival rates compared with those without early thrombocytopenia (17.3 vs. 7.6 months, respectively; P<0.001). Therefore, it was hypothesized that thrombocytopenia may be a predictive marker of TTD in treatment with T-DM1, considering that, in previous studies, the majority of instances of grade ≥3 thrombocytopenia had occurred within the first 42 days (4,26) and that 70% of total dose reductions due to adverse events had occurred within the first 4 months (27).

Experimental studies evaluating the potential mechanisms of thrombocytopenia have demonstrated that, rather than exerting a direct effect on platelets, T-DM1 inhibits the proliferation and differentiation of proplatelet precursors by disrupting microtubules after the uptake of megakaryocytes via micropinocytosis (28,29). Tang et al (15) hypothesized that emtansine molecules released from lysed HER2⁺ tumor cells after the administration of T-DM1 enter the systemic circulation and cause adverse events, such as hepatitis and thrombocytopenia, explaining the association between efficacy and adverse events. The findings of the present study and the high incidence of adverse effects in the first cycles of treatment, that is, during the period when the tumor burden was highest, are in agreement with this hypothesis, but it should be confirmed by further studies.

Treatment beyond radiological progression with targeted anticancer therapies, including tyrosine kinase inhibitors, anti-vascular endothelial growth factor and checkpoint inhibitors, has been studied in various solid organ malignancies, such as lung cancer, renal cell carcinoma and melanoma (30–37). These studies indicate that the continuation of these agents after progression could contribute to increased survival rates, especially in selected patients, and the significance of predictive markers in this regard has been observed. The unique adverse events of numerous targeted agents that occur during treatment have been associated with improved treatment responses and prolonged survival times (38–41). In this context, it is hypothesized that thrombocytopenia may be used to select patients to continue T-DM1 treatment in a post-progression setting after the predictive value is confirmed in large scale studies.

Other factors affecting TTD were the ECOG performance score and hormone receptor status. A low ECOG performance score was associated with improved survival outcomes, consistent with previous reports (16,42,43). It is hypothesized that the significantly shorter TTD of patients with a hormone receptor-positive status in the present study was associated with the preference of clinicians to discontinue T-DM1 and add endocrine therapy after radiological progression in these patients.

The main strength of the present study was the use of TTD instead of PFS as the endpoint. TTD was reported to inform the clinician regarding the continuation of an anticancer agent after objective progression, which is common in oncology practice, particularly for targeted agents, and has a high association with PFS, which is an objective evaluation (44). In addition, the present study included a relatively high number of patients compared with the aforementioned studies on the predictive value of thrombocytopenia. However, the present study also had limitations, including its retrospective design and the lack of toxicity data other than laboratory findings. Furthermore, factors impacting thrombocytopenia could not be analyzed due to a lack of associated data, such as pretreatment platelet counts, which have been identified as a strong predictor for thrombocytopenia during T-DM1 treatment (26). Moreover, tumor burden, one of the aspects affecting response in cancer treatments (45), could not be assessed due to the retrospective and multicenter design of the present study.

In conclusion, the patients that experienced thrombocytopenia in the first four cycles of T-DM1 treatment had a longer TTD compared with those without thrombocytopenia. Thus, future prospective studies and large-scale cohorts should aim to confirm the predictive role of thrombocytopenia for improved survival rates in order to maximize the potential benefit of T-DM1 treatment.