Join Date: Aug 2006
No Pain, No Gain: A Fallacy So Far
From the Journal of Clinical Oncology:
Corresponding author: William J. Gradisha, MD, Northwestern University, Department of Medicine, Division of Hematology/Oncology, 676 North Saint Clair Street, Suite 850, Chicago, IL 60611
The notion that toxicity related to treatment can be a surrogate for treatment efficacy has long been embedded in the psyche of patients. We are frequently confronted with the question of whether the therapy we have recommended is aggressive—not infrequently a path the patient desires, even if it means more adverse events. Oncologists have also often been attracted to this approach to treatment, believing that a bad disease needs to be treated aggressively, and again this often relates to the adverse events that a patient may experience. However, this philosophy has not always proven to be correct. For example, in the realm of breast cancer therapeutics, extensive radical surgeries (eg, Halsted radical mastectomies and regional nodal dissections) have been largely abandoned as evidence from randomized trials revealed that far less breast surgery, radiation therapy, and minimal axillary nodal sampling results in equivalent clinical outcomes: breast preservation with far less morbidity. Similarly, not so long ago, the enthusiasm for high-dose chemotherapy and autologous stem-cell transplantation as treatment for both early-stage and advanced-stage breast cancer led to a nationwide proliferation of programs offering such treatments for patients. The conviction that this form of therapy was so much more effective than standard systemic therapy delayed the completion of randomized clinical trials by many years; and those trials ultimately proved that high-dose chemotherapy did not offer an advantage over standard-dose systemic therapy. Multiple other examples in breast cancer and other malignancies can be offered as evidence that toxicity of a therapy does not necessarily correspond in a meaningful way to the likelihood of clinical benefit. The article by Stearns et al1 that accompanies this editorial shows no relationship between the effectiveness of adjuvant anastrozole and the development of either vasomotor or musculoskeletal symptoms.
Much of the toxicity that systemic therapy confers to a patient is the result of collateral damage—nontargeted effects to normal tissue that may, or may not, share specific features of the target cancer cell. The idea that specific toxicities from an anticancer agent may be a surrogate for efficacy has gained traction in recent years. For example, an acne-form skin rash is an extremely common adverse effect of epidermal growth factor receptor (EGFR) tyrosine kinase inhibitors (TKI), such as gefitinib or erlotinib, occurring in up to 66% of patients with non–small-cell lung cancer (NSCLC). A meta-analysis of clinical trials evaluating EGFR-TKIs as treatment for NSCLC have shown that the development of a skin rash is a marker for response to the EGFR-TKI and also predictive of better progression-free survival and overall survival compared with those who do not develop such a rash. EGFR is important in both skin biology and NSCLC.2 EGFR-TKIs seem to interact with basal keratinocytes, which results in the rash that may be an outward manifestation of a similar effect occurring in tumor tissue.
The use of tamoxifen as adjuvant therapy for early-stage breast cancer can now be recommended as monotherapy for 5 years, for 2 to 3 years to be followed by an aromatase inhibitor (AI) to complete 5 years, or as long as 10 years in women deemed to have a significant ongoing risk of recurrence who remain premenopausal or are not candidates for treatment with an aromatase inhibitor. Any medication that must be taken daily for years can be a challenge for both the patient and the physician. Extensive research has shown that compliance is a major issue with longer durations of therapy and that rates of compliance are influenced by many factors including age, education, and the magnitude of the insurance copay that a patient must contribute.3–5 Although physicians have reported confidence in medication compliance by their patients, evidence suggests otherwise.3
Hot flushes have been suggested as a surrogate for tamoxifen efficacy. The underlying explanation purported to support this observation is that women who are more able to metabolize tamoxifen to its active metabolite, endoxifen, will be those experiencing more hot flushes. By extension, those women unable to metabolize tamoxifen as effectively, resulting in less endoxifen and fewer hot flushes, would be those predicted to have less benefit from tamoxifen and greater numbers of recurrences. The basic concept was that occurrence of hot flushes was hypothesized to predict which patients were likely to be extensive metabolizers (EM) compared with those less able to metabolize tamoxifen (intermediate metabolizers [IM] or poor metabolizers [PM]). In the Women's Healthy Eating and Living (WHEL) study, a nutritional intervention study, Mortimer et al6 reported on 1,551 patients with early-stage breast cancer who were enrolled and randomly assigned to the no dietary intervention group. At baseline, 56% of these women were taking tamoxifen, and of those, 78% reported hot flushes. After 7.3 years of follow-up, patients who reported hot flushes at baseline were less likely to develop a recurrence (12.9% v 21%) compared with those without hot flushes at baseline. This observation was a stronger predictor of recurrence compared with patient age, hormone receptor status, or disease stage. Although no one was eager to tell a woman hot flushes were a good thing, these results suggested that perhaps the occurrence of hot flushes could be a poor man's assay for CYP2D6 genotype.
However, CYP2D6 testing has had its share of controversy. Despite early results that suggested the ability of CYP2D6 testing to predict which patients were most likely to benefit from tamoxifen (EM), subsequent work on several large, mature data sets have failed to confirm the early findings, and most show no association between CYP2D6 metabolic phenotype and clinical outcome of patients on tamoxifen. Regan et al7 reported on 4,861 of 8,010 patients who participated in the Breast International Group (BIG) 1-98 study, which randomly assigned patients to tamoxifen for 5 years, letrozole for 5 years, or a sequence of tamoxifen (2 years) followed by letrozole (3 years) or letrozole (2 years) followed by tamoxifen (3 years). DNA from tumor tissue was archived from patients participating in the study. Subsequently, CYP2D6 genotyping categorized the patients as EM, IM, and PM, but no association between CYP2D6 metabolism genotypes and breast cancer–free interval in those patients receiving tamoxifen was observed. Even more puzzling and contrary to the hypothesis, the patients with PM and IM phenotypes reported an increased rate of tamoxifen-induced hot flushes. These data and other reports undermine any potential role for CYP2D6 testing in patient management, and current practice guidelines do not support use of CYP2D6 testing to make treatment decisions regarding the use of tamoxifen.8
As another example of the argument “no pain, no gain,” analyses have been conducted in patients with early-stage breast cancer receiving adjuvant chemotherapy containing a taxane to determine whether patients developing peripheral neuropathy were more likely to benefit from treatment. Moreno-Aspitia et al9 retrospectively reported on the experience of patients with human epidermal growth factor receptor 2–positive, early-stage breast cancer enrolled onto the Breast Intergroup Trial N9831 who were randomly assigned to doxorubicin/cyclophosphamide followed by: (1) 12 weeks of paclitaxel, (2) 12 weeks of paclitaxel followed by trastuzumab, or (3) 12 weeks of paclitaxel with concurrent trastuzumab. Only patients who initiated paclitaxel and did not have peripheral neuropathy at the initiation of paclitaxel were included in the analysis; of 1,844 eligible patients in arms A and B, 379 (20.9%) developed peripheral neuropathy. For patients in arm A (without trastuzumab), those who developed peripheral neuropathy had better disease-free survival (DFS) than those patients who did not (3-year DFS, 86.2% v 81.8%; P = .01) despite dose reductions of paclitaxel in those with neuropathy. In patients receiving trastuzumab (arm C), there was no difference in outcome (DFS) in those with or without peripheral neuropathy. It was suggested that the onset of neuropathy could represent the target effect of the taxane binding to tubulin, whereas the added benefit of trastuzumab in arm C may have overshadowed the tubulin binding effect of paclitaxel. However, another secondary analyses by Schneider et al10 in patients with early-stage breast cancer participating in Eastern Cooperative Oncology Group Study 1199 who were randomly assigned to doxorubicin/cyclophosphamide followed by four different taxane schedules (paclitaxel 175 mg/m2 once every 3 weeks for four weeks or once per week at 80 mg/m2 for 12 weeks; docetaxel 100 mg/m2 every 3 weeks for four weeks or once per week at 35 mg/m2 for 12 weeks) showed no association with therapeutic benefit. Although 770 patients (17% of the total) experienced peripheral neuropathy (grade 2-4) after adjuvant taxane therapy with the highest rate among those receiving paclitaxel once per week, there were no significant associations between the development of peripheral neuropathy and any clinical end points including DFS, recurrence-free survival, or overall survival.
Vasomotor symptoms and musculoskeletal complaints are well-recognized adverse events associated with AIs. With the exception of postmenopausal women unable to tolerate an AI, AI therapy has become an integral part of adjuvant endocrine therapy for postmenopausal women with early-stage breast cancer given either as monotherapy or in sequence with tamoxifen (BIG 1-98, Arimidex, Tamoxifen Alone or in Combination [ATAC], Tamoxifen Exemestane Adjuvant Multinational [TEAM], etc). Long durations of oral therapy alone are associated with noncompliance even in the absence of significant adverse events.3 Postmenopausal women who have severe joint symptoms while taking an AI are even more likely to discontinue treatment. Chim et al11 recently reported on a retrospective cohort study of postmenopausal women taking an AI using a survey instrument focusing on symptoms with special attention to specific measurement of joint pain. Among 437 eligible patients, 11% prematurely discontinued AI therapy an average of 29 months after starting. The most common reason for premature discontinuation of the AI was joint pain (57%) as assessed by the Brief Pain Inventory. The question then emerges as to whether treatment-related musculoskeletal complaints may predict those patients likely to derive benefit from an AI.
The ATAC investigators reported that 55% of patients receiving either monotherapy with tamoxifen or anastrozole developed vasomotor or musculoskeletal symptoms within 3 months of starting therapy and were also those found to have superior relapse-free survival (RFS).12 Despite the symptoms, adherence was remarkably high, which could have resulted from the allowed use of nonsteroidal anti-inflammatory drugs and COX2 inhibitors. The TEAM trial randomly assigned 9,766 patients to 5 years of exemestane or sequential treatment with tamoxifen followed by exemestane for a total of 5 years.13 In an analysis of specific adverse events in TEAM and their relationship to survival benefit, Fontein et al14 reported that patients who experienced musculoskeletal complaints or vulvovaginal symptoms were more likely to have a better DFS, whereas only those with vulvovaginal symptoms had better overall survival. This analysis also has potential limitations including lack of information about pre-existing symptoms, reporting of symptoms occurring only in the first year, and patient self-reporting as the source of data rather than a defined instrument for gathering the information. The concurrent use of nonsteroidal anti-inflammatory drugs was also not reported.
Stearns et al1 report on treatment-related musculoskeletal and vasomotor symptoms observed in trial NCIC MA.27. This study, which randomly assigned 7,576 postmenopausal women with early-stage breast cancer to tamoxifen or anastrozole for 5 years and, different than other studies, patient-reported symptoms were collected using the standardized Common Terminology Criteria for Adverse Events (CTCAE) version 3.0 at baseline, 6 months, and 12 months. Symptoms were defined as present, if either vasomotor and/or joint complaints were reported at any of the specified time points. Ninety-six percent of patients were included at both the 6- and 12-month assessments. Thirty-four percent had baseline symptoms. For the remaining 66% of patients without baseline symptoms, 25% had new symptoms at 6 months, and 53% had them by 12 months. The MA.27 investigators found no relationship between new or worsening vasomotor or joint symptoms on RFS, even when accounting for the use of celecoxib, aspirin, trastuzumab, and chemotherapy exposure. However, the high 4-year event-free survival in MA.27 (more than 90% for the anastrozole and the exemestane groups) would make it difficult to demonstrate a difference in recurrence-free survival for this secondary analyses, should one exist.
Therefore, perhaps it is time for us to be fair to ourselves and to our patients and recalibrate our thinking. Toxicity must not be confused as an automatic surrogate measure of clinical benefit. As the number of clinical options expands for patients with cancer, and breast cancer in particular, the focus may be more appropriately placed on understanding which patients are most likely to develop a specific toxicity. Can we identify biomarkers (ie, polymporphisms) that accurately identify those patients at greatest risk for a particular toxicity? If we have alternative but equal treatment options or ways of adjusting dosing in an individual patient, we may be able to not only provide targeted cancer therapy but also provide the precision treatment of individual patients we all hope for.