Enhancing Therapeutic Decision Making When Options Abound: Toxicities Matter

Medical oncology has come a long way since early chemotherapy studies for advanced-stage cancers. For most malignancies, we now have multiple treatment options to discuss with our patients. Many patients, in particular, the majority of those diagnosed with early-stage breast cancer, have an excellent chance of survival and should go on to live full lives. At the same time, the clinical decision-making process in early-stage breast cancer has become more complicated because clinicians and patients must navigate through various chemotherapy options that might offer similar reductions in cancer recurrence and death. Consequently, considerations beyond treatment efficacy, particularly the possibility of serious and life-threatening shortand long-term toxicities, assume even greater importance in this decision-making process. Older studies revealed that patients were willing to accept greater risks of toxicity in scenarios where chemotherapy offered as little as a 1% absolute survival improvement. Our evolving abilities to identify patients who attain only limited benefit from chemotherapy allow us to be more selective about when to recommend more toxic options. At the same time, individual patient preferences and quality-of-life considerations can vary considerably. Yet, in a curable setting, intuitively, patients and physicians alike consider iatrogenic death early in the disease course to be particularly tragic; these are also described as situations of high regret because of the many years of life prematurely lost. A number of chemotherapy complications can be fatal. Infections in the setting of myelosuppression and thromboembolic events remain common causes of early death as a result of chemotherapy across a range of solid tumors and lymphoma. Hospitalizations during or immediately after adjuvant therapy represent an indirect but important surrogate for early, potentially life-threatening complications. In contrast to efficacy results from randomized clinical trials (RCTs) in highly selected patients, effectiveness studies aim to assess how cancer treatments perform in the unselected, frequently sicker, general cancer population. These clinical factors can adversely impact the benefit of therapies observed and deemed safe within the context of healthier patients in RCTs. Surprisingly, there are limited effectiveness data available on the impact of acute and life-threatening toxicities from specific breast cancer therapies. Even when most study participants in RCTs are recruited from community settings, the reported toxicity rates may not necessarily reflect their true incidence in the general patient population. An important recent example in breast cancer was the discrepancy that was observed between the frequency of serious infectious toxicities seen with docetaxel and cyclophosphamide (TC) in the original publication compared with subsequent cohort studies in unselected patient populations. Another recent breast cancer study observed unsuspected inappropriate prophylactic antibiotic use in US oncology outpatient practice, whereas Hassett et al reported higher than expected rates of chemotherapy-associated hospitalizations (12%) and toxicities in younger patients with breast cancer. Likewise, recent randomized trial data suggest that there remains uncertainty about the burden of neuropathy associated with various microtubule inhibitors and drug schedules. The traditional reliance on provider clinic notes as source documentation, which are then transcribed by study coordinators into case report forms, further increases the opportunity for recording errors and underreporting of the true burden of toxicities. If the patient fails to mention a particular symptom, the associated adverse effect might well not be recorded, even in high-quality RCTs. Inconsistent data collection together with unknown variations in supportive care use likely explain the remarkable variability of hematologic toxicities reported in numerous, otherwise similar RCTs that use the exact same treatment regimen. Such reporting variability would be inconceivable for primary efficacy outcomes assessment. Prospective cohort studies performed in real-world settings and designed specifically to assess toxicities with more consistent data collection of adverse events enhance our knowledge base. Unfortunately, such prospective cohort studies are relatively costly and time consuming and have been provided with limited funding opportunities. It is with these issues in mind that we comment on the article by Barcenas et al that accompanies this editorial. The authors undertook an important retrospective study in a general breast cancer population diagnosed between 2003 and 2007. They assessed the frequency of chemotherapy-related hospitalizations among six major adjuvant chemotherapy regimens within 6 months of starting therapy: TC, doxorubicin and cyclophosphamide (AC), TAC, AC plus T (AC T), dose-dense AC plus paclitaxel (ddAC P), or AC plus weekly P (AC wP). They used the SEER/Texas Cancer Registry Medicare (SEER-Medicare) claims database to assess patients older than 65 years of age and the MarketScan administrative data set for those younger than 65 years of age. The investigators report that crude hospitalization rates within 6 months of the start of therapy in patients younger than age 65 years JOURNAL OF CLINICAL ONCOLOGY E D I T O R I A L VOLUME 32 NUMBER 19 JULY 1 2014