Our survey provides new insight into the implementation of adjuvant therapies for HER2+ breast cancer. This survey captured a spectrum of treatment preference, ranging from oncologists who exclusively select TCH to those who use only anthracycline-based therapies. However, we also identified that a large number of oncologists take a more nuanced approach when selecting regimens, using TCH for small node-negative tumors and AC-TH for advanced-stage tumors where the additional risk of cardiotoxicity is overshadowed by risk of cancer recurrence. Moreover, we found the dose/schedule and duration of trastuzumab is frequently altered for a myriad of reasons. To our knowledge randomized studies are not planned to determine which of these treatment approaches is optimal, making analysis of patient outcomes important for optimizing current care.
As expected, cardiotoxicity is a major concern among oncologists. Cardiotoxicity of trastuzumab-based therapies has been extensively scrutinized [3
]. Although cardiotoxicity is important, our study signals a note of caution regarding the importance of other toxicities. In particular, acute myelosuppression is more common with TCH compared with AC-TH, resulting in higher rates of febrile neutropenia, anemia, thrombocytopenia, and consequent hospitalizations. We conclude that, in addition to cardiotoxicity, myelosuppression should be considered when selecting TCH vs. AC-TH.
The perception amongst the surveyed oncologists is that TCH and anthracycline-based regimens are equally likely to be completed without delay or modification. In practice, however, we found a higher rate of dose delays and modifications with TCH than AC-TH. The major toxicity was myelosuppression and neutropenic fever with 26% risk without prophylactic G-CSF. This result seems dissonant with phase III results, where TCH is reported to have a 66% risk of grade 3/4 neutropenia, but only an 9.6% rate of febrile neutropenia[3
]. It is unlikely that our population was at greater risk of myelosuppression, since half of hospitalizations were in healthy women <50 years old. However, we note that Slamon et al. reported that G-CSF use was left to the discretion of the treating oncologist, but does not specify how often this was used [3
]. Moreover, these investigators report an additional 11.2% risk of grade 3/4 neutropenic infection with TCH. Using Common Terminology Criteria for Adverse Events version 2.0, neutropenic fever and neutropenic infection categories are distinguished by whether an infectious organism was identified, so total rate of neutropenic fever plus infection is 20.8% in BCIRG 006, despite use of G-CSF in some patients. Interestingly, more than half of Wisconsin oncologists who were surveyed report using prophylactic G-CSF with TCH, suggesting that oncologists recognize this toxicity through individual experience. Based on the rates of neutropenic fever/infection and current guidelines [19
], we conclude that prophylactic G-CSF should be used routinely with TCH.
These findings demonstrate the importance of clear reporting for GCSF use in large phase III trials of adjuvant therapy. Previously, US Oncology 9735 reported a 5% rate of febrile neutropenia with docetaxel and cyclophosphamide, without specifying rate of G-CSF use. However, two observational studies reported rates of 33-50% with this regimen when G-CSF was not used [20
]. Similarly, we report an unexpectedly high rate of febrile neutropenia with TCH. We conclude that phase III studies of chemotherapy regimens should carefully record and report G-CSF use among subjects, to allow oncologists to assess myelosuppression and to safely apply results to patients.
We anticipate possible increases in myelosuppression due to TCH with implementation of Isotope Dilution Mass Spectrometry (IDMS) for measurement of creatinine. IDMS has recently been adopted as a calibrated measure for creatinine, with values that are typically lower than what was determined previously by individual laboratories [22
]. With TCH, carboplatin is dosed using AUC 6 [3
], typically calculated Creatinine Clearance using the Calvert formula[23
]. Both UW and Marshfield implemented the IDMS measurements after all patients reported here were treated. If IDMS was used, most patients would have received higher doses of carboplatin, making the likelihood of neutropenic fever and myelosuppression even higher than we report here.
When implemented in routine practice, HER2+ regimens are commonly modified from the schedules used in phase III trials. Both the NSABP trial B-31 and the NCCTG trial N9831 gave AC every 3 weeks [2
], yet in our study, AC was most commonly given in a dose-dense fashion, every two weeks, followed by paclitaxel. Although this schedule differs from phase III studies, dose-dense AC is not associated with an increase risk of clinical cardiotoxicity or LVEF decline at 6 months, even when combined with HER2-targeted therapy [15
]. We found that oncologists also routinely modify TCH schedule to use trastuzumab every 3 weeks; we did not observe unexpected toxicity associated with this modification.
Strengths of this study include a high rate of response among surveyed physicians and the inclusion of a population of patients seen in routine practice. Moreover, the variations in clinical practice in this study allowed correlation of interventions and toxicity (e.g. G-CSF and febrile neutropenia). Finally, AC-TH and TCH cohorts are well matched for factors other than tumor stage. A major limitation of the chart review is that it is retrospective with limited objective endpoints, non-randomized and limited to two regional oncology centers. Despite the limitations, retrospective analysis can provide important guidance for selecting treatment and supportive care given the lack of planned prospective studies to evaluate these questions.
In conclusion, this study demonstrates wide variation in perception and practice among oncologists who select therapy for HER2+ breast cancer. We conclude that a major focus on cardiotoxicity has obscured other differences between anthracycline- and non-anthracycline based treatments, even though all toxicities should be considered when selecting a cancer treatment regimen. In particular, we find that myelosuppression with TCH is more severe than with AC-TH and meets guidelines for routine use of G-CSF.