Preoperative Chemotherapy for Operable Breast Cancer Is Associated with Better Compliance with Adjuvant Therapy in Matched Stage II and IIIA Patients

Patient compliance with physician recommendations for all types of adjuvant therapy for operable breast cancer in the community setting was examined.

Abstract

Introduction.

Preoperative chemotherapy (PC) for operable breast cancer has shown significant benefits in prospective trials. Many patients are treated in the community setting and some may question the applicability of PC outside the university setting.

Methods.

Retrospective review was performed of stage II and IIIA breast cancer patients treated from January 2002 to July 2009. Fifty-three of 57 patients who underwent PC were matched based on age, tumor size, and hormone receptor status with 53 patients who did not undergo PC. Differences in patient compliance with physician recommendations for all types of adjuvant therapy were evaluated. Crude odds ratios and adjusted odds ratios derived from conditional logistic regression models were calculated.

Results.

There were 106 patients included. Patient compliance with chemotherapy was better in the PC group than in the adjuvant chemotherapy (AC) group (100% versus 70%; p = .0001). Similarly, more patients in the PC group completed radiation therapy (96% versus 65%; p = .0003) and initiated hormonal therapy (100% versus 62%; p = .0001).

Conditional logistic regression revealed that higher pathologic stage and current cigarette smoking were associated with poorer compliance with chemotherapy. For radiation therapy, the univariate model revealed that compliance with chemotherapy and being employed were associated with completion of radiation, whereas current cigarette smoking and larger pathologic size were associated with poorer compliance with radiation. For hormonal therapy, current cigarette smokers were more likely to be noncompliant with initiation of hormonal therapy.

Conclusions.

PC for operable breast cancer can improve patient compliance with chemotherapy. Current cigarette smokers were more likely to be noncompliant with all types of adjuvant therapy.

Introduction

Patients treated with preoperative chemotherapy (PC) for operable breast cancer were shown to have a similar survival time to those treated with adjuvant chemotherapy (AC) in prospective trials [1–4]. Despite the lack of difference in survival outcomes, there were benefits recognized in patients who received PC. First, PC allows an in vivo test of chemosensitivity of the cancer. Tumor response as measured by the postchemotherapy pathologic tumor burden in the breast and lymph nodes correlates with overall survival [3–5]. In addition, for patients who do not respond to the chemotherapeutic regimen, optimization of therapy can be obtained by terminating the regimen and changing to a different regimen or type of treatment [4, 6]. In patients undergoing AC, this type of adjustment is not possible because there is no reliable clinical marker for tumor response once the primary tumor and lymph nodes have been removed.

In the National Surgical Adjuvant Breast and Bowel Project (NSABP) B18 and NSABP B27 studies, PC led to a higher rate of lumpectomy and higher rate of pathologically negative lymph nodes than in patients who underwent AC [1, 2]. The higher rate of lumpectomy was achieved without a significantly greater risk for local recurrence [1–4]. In patients with tumors ≥3 cm in size who underwent PC, Bonadonna et al. [3] demonstrated a 6.8% risk for local recurrence at 65 months of follow-up. In addition to the clinical benefits, PC is a powerful research tool because it allows collection of “before and after” specimens for evaluation. The potential endpoint represented by pathologic response allows prospective studies to be performed in a shorter time period and with fewer patients [7, 8]. These benefits have allowed the use of PC to become common in the clinical trial setting and at large referral cancer centers. Surgical intervention, however, remains the most common initial treatment for operable breast cancer.

One possible reason for the infrequent use of PC is the question of its efficacy outside the clinical trial or away from the university setting. Many patients are treated in the community and some physicians may question the use and benefits of PC and its applicability in the “nonclinical trial” patient care practice. Some practitioners may believe that because overall survival is the same whether chemotherapy is given preoperatively or in the adjuvant setting, the only clinical indication for PC is to give a patient with a large cancer the possibility of conserving her breast. In modern breast cancer treatment, with better technology constantly allowing us to better determine which patients benefit from chemotherapy, PC can be applied to a more appropriate and carefully selected patient population [9–12].

We sought to examine the applicability of and potential benefits to the use of PC for operable breast cancer in the community setting. Patient compliance with physician recommendations for all types of adjuvant therapy was examined.

Materials and Methods

Maricopa Medical Center is a 679-bed facility that is the county, safety-net hospital in Phoenix, Arizona, and serves a patient base that is 35% non–English speaking. The majority of the patients treated are underinsured. Underinsured patients were considered to be all patients without commercial insurance. In the state of Arizona, these patients are covered by Medicare and the AHCCCS (Arizona Health Care Cost Containment System, Medicaid). Patients ineligible for AHCCCS were classified as uninsured self-pay or charity care. Institutional review board approval was obtained.

A retrospective review was performed of all breast cancer patients with a primary breast cancer from January 2002 to July 2009. Only patients who presented with American Joint Committee on Cancer (AJCC) clinical stage IIA, IIB, or IIIA breast cancer were included in the study. All patients having their primary operation prior to January 2002 were excluded. Patients presenting with AJCC clinical stage 0, I, IIIB, or IV breast cancers were excluded. Other criteria excluded all male patients and all patients with a second primary or recurrence who had the primary breast cancer prior to January 2002. After exclusions, a total of 212 patients underwent treatment during this time period. Fifty-seven patients were in the PC group and 155 were in the AC group. Because patient age at diagnosis, clinical tumor size, and hormone receptor status are factors known to affect physician recommendations for AC and PC, patients in this study population were matched based on these variables. Fifty-three of the 57 PC patients were matched with 53 AC patients. Four patients from the PC group could not be matched because their young age precluded identification of a counterpart in the AC group. Sociodemographic, clinical, and histopathologic characteristics for patients were collected.

Chemotherapy

Breast cancer treatment is very individualized, and determining parameters for categorization in a study is controversial. Although there are always exceptions in clinical practice, because the current study included patients with at least clinical stage IIA breast cancer, all patients were initially considered candidates for chemotherapy. Determination was then made of patients who did not require chemotherapy. In alignment with the National Comprehensive Cancer Network (NCCN) guidelines for clinical practice [13], any patient aged >70 years was excluded from the analysis of compliance. Patients with a low or intermediate recurrence score on the Oncotype DX® assay (Genomic Health, Redwood City, CA) were considered not required to have chemotherapy [10]. Finally, according to documentation in the chart, if the medical oncologist did not recommend chemotherapy for any reason (i.e., comorbidities, etc.), then the patient was excluded from the analysis of compliance. Patient compliance with physician-recommended chemotherapy was therefore evaluated in patients considered to be “candidates” for chemotherapy by the above criteria. If the oncologist recommended chemotherapy and the patient did undergo the chemotherapy, the patient was counted as compliant. The number of cycles of each chemotherapeutic regimen was documented. If the patient did not initiate the treatment, then the patient was counted as noncompliant.

If patients were candidates for chemotherapy by the above criteria and presented with a clinically palpable cancer ≥2 cm in dimension, consideration for PC was given by the surgical oncologist or medical oncologist. PC was particularly discussed if the patient was interested in but not an appropriate candidate for a breast-conserving operation at the time of presentation. For all patients, the chemotherapeutic regimen was the choice of the treating oncologist. In the PC group, the regimens included doxorubicin and cyclophosphamide (11%), doxorubicin and cyclophosphamide followed by paclitaxel (51%), doxorubicin and cyclophosphamide followed by docetaxel (14%), docetaxel, doxorubicin, and cyclophosphamide (9%), trastuzumab-containing regimens (12%), and other regimens (4%).

Radiation Therapy

All patients undergoing breast-conserving operations were recommended to undergo radiation therapy. In patients who had a mastectomy, all patients who had an invasive cancer with pathologic tumor size ≥5 cm or had four or more lymph nodes involved with cancer were recommended to have postmastectomy radiation therapy. For the purposes of reporting, because N1 (one to three positive lymph nodes) disease is felt to be a controversial indication for postmastectomy radiation by many physicians, these patients were considered “not required” to undergo postmastectomy radiation therapy.

Hormonal Therapy

All patients with cancers that expressed hormone receptors (at least 10% by immunohistochemistry) were considered candidates for hormone therapy. In this study, the only exception was for those patients with comorbidities considered a contraindication as documented by the medical oncologist.

Statistical Considerations

The study population was classified into two groups based on whether the patient received PC or was expected to have AC by the treating physician. Fifty-three of the 57 PC patients were matched according to patient age, tumor size, and hormone receptor status with 53 patients from the AC group. Differences in sociodemographic characteristics, clinical measures, and outcomes of breast cancer care were evaluated in the study population. For categorical variables, the Fisher's exact test for independence was used to determine differences between the two populations. For each continuous variable, for example, age, body mass index (BMI), etc., the mean and standard deviation were reported. Unpaired (two-sample) t-tests were used to compare the means of the two populations. All statistical tests were two-sided and a p-value ≤ .05 was considered statistically significant. Univariate logistic regression analysis was performed to assess variables that may affect patient compliance with physician recommendations for each type of adjuvant therapy. The matching variables, patient age at diagnosis, tumor size, and hormone receptor status, were not included in the analysis. A conditional rather than unconditional logistic regression analysis was then performed to account for the effects of matching and derive adjusted odds ratios. Both crude odds ratios (ORs) and adjusted ORs derived from univariate unconditional logistic regression models and conditional logistic regression models, respectively, were calculated along with their associated 95% confidence intervals (CIs) and p-values to study the association between patient compliance with physician recommendations for chemotherapy, radiation therapy, and hormone therapy and each sociodemographic and clinical variable. Previous studies have demonstrated that distance from a radiation oncology facility correlates with patient compliance with radiation therapy [14–16]. This factor was not included in the analysis because nearly all the patients treated in the current study live in the Phoenix metropolitan area and do not live >25 miles from a radiation oncology facility.

In addition, to test the hypothesis that chemotherapy is the lynchpin for completing all recommended adjuvant therapies, completion of chemotherapy was included as a variable in the assessment of compliance with radiation and hormone therapy.

Results

There were 106 patients included in the matched analysis. Sociodemographic information is shown in Table 1. The patients in the two groups were very closely matched with respect to age at diagnosis, as expected. Patients in the PC group were more likely to be employed than those in the AC group; however, there was no difference in insurance status. Most patients in both groups were Hispanic and insured by AHCCCS (Medicaid) or uninsured. Very few patients in either group had undergone a screening mammogram within 2 years of presentation (Table 1). With respect to clinical characteristics, patients were matched for clinical tumor size and hormone receptor status (Table 2). The histopathologic factors were similar with respect to predominant histology, histologic grade, human epidermal growth factor receptor 2/neu status, and proportion of triple-negative breast cancers. There was a high rate of poorly differentiated tumors in both groups.

Table 1.

Sociodemographic information for matched stage IIA, IIB, and IIIA breast cancer patients

^aDerived from a Fisher's exact test for categorical variables and a two-sample t-test for continuous variables.

Abbreviations: BMI, body mass index; SD, standard deviation.

Table 2.

Clinical and histopathologic data for matched stage II and IIIA breast cancer patients

^aDerived from a Fisher's exact test for categorical variables and a two-sample t-test for continuous variables.

^bFive patients in the adjuvant chemotherapy group and six patients in the preoperative chemotherapy group did not have a histologic grade assigned either because it was an ILC or not done.

^cHormone receptor and HER-2/neu not expressed/not amplified.

Abbreviations: FISH, fluorescence in situ hybridization; HER-2, human epidermal growth factor receptor 2; IHC, immunohistochemistry; ILC, infiltrating lobular carcinoma; ND, not done; SD, standard deviation.

PC resulted in a clinical response in 80% of patients. Of these, 55% had a complete response and 25% had a partial response. The remaining 20% had either stable disease or disease progression. Overall 26% of patients had a pathologic complete response.

Patient compliance with physician-recommended adjuvant therapy was significantly better in the PC group than in the AC group. All patients who initiated PC completed their treatment. Four (8%) patients, however, did not complete one or two cycles prior to operation. Three of the four patients completed the regimen after the operation. Information on dose reductions and dose delays, however, was not collected. For those candidates in the AC group without documented contraindications, only 70% initiated the chemotherapeutic regimen recommended by their physician (p = .0001). Of those who did initiate chemotherapy in the AC group, one (2%) patient did not complete three cycles of their planned regimen. Similarly, more patients in the PC group completed radiation therapy (96% versus 65%; p = .0003) and initiated hormonal therapy (100% versus 62%; p = .0001) than in the AC group.

Evaluation of variables associated with patient compliance with physician recommendations for each type of adjuvant therapy was then performed. Sociodemographic and treatment-related variables were included as well as the more commonly considered clinical variables. For compliance with chemotherapy, univariate analysis revealed that patients with a lower BMI and smaller pathologic tumor size as well as those having undergone PC had better compliance (Table 3). Patients who were current smokers were more likely to be noncompliant with physician recommendations for chemotherapy. The adjusted analysis, which was a conditional logistic regression to take into account the effects of matching, revealed that lower pathologic stage was associated with better patient compliance and current cigarette use was associated with significantly poorer patient compliance with chemotherapy.

Table 3.

Predictors of patient compliance with physician-recommended chemotherapy

^aDerived from a Fisher's exact test for categorical variables and a univariate logistic regression model for continuous variables.

^bDerived from a conditional logistic regression model that is used to eliminate the matching effect (age, clinical size, and estrogen receptor status).

Abbreviations: BMI, body mass index; CI, confidence interval; HER-2, human epidermal growth factor receptor 2; NA, not available; OR, odds ratio; SD, standard deviation.

For compliance with radiation therapy, current cigarette use was statistically significantly associated with noncompliance with completion of radiation therapy. Current employment and smaller pathologic size were associated with better compliance with radiation therapy. The strongest predictor of compliance with physician-recommended radiation therapy was whether or not the patient was compliant with chemotherapy (OR, 86.67; 95% CI, 9.21–814.7; p < .0001) (Table 4). For compliance with hormonal therapy, only current cigarette use was associated with a higher likelihood of not initiating hormonal therapy (OR, 0.16; 95% CI, 0.04–0.70; p < .0001) (Table 5).

Table 4.

Predictors of patient compliance with physician-recommended radiation therapy

^aDerived from a Fisher's exact test for categorical variables and a univariate logistic regression model for continuous variables.

^bDerived from a conditional logistic regression model that is used to eliminate the matching effect (age, clinical size, and estrogen receptor status).

Abbreviations: BMI, body mass index; CI, confidence interval; HER-2, human epidermal growth factor receptor; NA, not available; OR, odds ratio; SD, standard deviation.

Table 5.

Predictors of patient compliance with physician recommendations for adjuvant hormonal therapy

^aDerived from a Fisher's exact test for categorical variables and a univariate logistic regression model for continuous variables.

^bDerived from a conditional logistic regression model that is used to eliminate the matching effect (age, clinical size, and estrogen receptor status).

Abbreviations: BMI, body mass index; CI, confidence interval; NA, not available; OR, odds ratio; SD, standard deviation.

Discussion

In the current study, matched patients with stage IIA, IIB, and IIIA operable breast cancers who underwent PC appeared to have better compliance with physician-recommended chemotherapy. In addition, patients who underwent chemotherapy appeared to demonstrate better compliance with physician recommendations for adjuvant radiation therapy than patients who were noncompliant with the recommendation for chemotherapy.

Few previous studies have evaluated compliance of patients with physician recommendations for chemotherapy, and data that examine racial/ethnicity minority populations are more limited [17–21]. The reasons for a lack of compliance or underuse of adjuvant therapies are not well documented. Because patient age at diagnosis, tumor size, and hormone receptor status are important clinical factors that physicians use to recommend chemotherapy, these variables were used in the matching of the two groups. With respect to patient compliance with physician recommendations for chemotherapy, univariate analysis revealed that a lower BMI, smaller pathologic tumor size, and PC were associated with better patient compliance. The influence of BMI on compliance with chemotherapy is not clear. Some studies have demonstrated that patients with a high BMI have poorer outcomes after chemotherapy [22, 23]. The finding that current cigarette smoking was associated with poorer compliance with chemotherapy, however, may provide some insight into the patient population studied. Because obesity and the use of tobacco are well-known health concerns [24, 25], patients who display these characteristics may be more likely to also go against physician recommendations and not comply with their expected chemotherapeutic regimen. The negative association of being a current smoker with compliance with chemotherapy for breast cancer has not been previously reported. In other areas of medicine, however, being a smoker has been reported to be a predictor of noncompliance with medical advice [26–28].

The finding of better compliance in patients with a smaller pathologic tumor size on univariate analysis and lower pathologic stage on conditional logistic regression may initially seem surprising. The effect of PC on compliance, however, can explain this result. Further analysis of the two groups revealed that pathologic size and stage were significantly associated with use of PC, as would be expected based on the well-known downstaging effect of PC [1–3]. A Wilcoxon rank-sum test showed that patients in the PC group had a significantly lower pathologic stage than those in the AC group (p = .0001).

Bickell et al. [21] attempted to evaluate the reasons for underuse of adjuvant therapies. Thirty-one percent of patients declined physician-recommended adjuvant therapy. Factors associated with patient compliance were not identified but race, age, insurance, and the physician's perception of the patient's social support were not related to declining recommended adjuvant therapy. These data are similar to the data from patients in the AC group of the current study, 30% of whom declined physician-recommended chemotherapy. At Indiana University [19], a similar difference in patient compliance with physician recommendations for chemotherapy was seen in a county hospital population (PC group, 100%; AC group, 73%; p = .01).

Patients in the PC group demonstrated better compliance with completion of radiation therapy. In the AC group, 65% of patients completed the recommended radiation, which is significantly lower (p = .0003) than the 96% of patients who completed radiation in the PC group (Table 3). This rate of compliance in the PC group compares favorably with that achieved at NCCN designated centers [29]. In the univariate model, current employment was significantly associated with completion of radiation. The reason for the effect of employment was not a reflection of insurance status because this was not found to be a significant predictor of compliance. At Wishard Hospital, the county hospital affiliated with Indiana University, current employment was statistically significantly associated with better breast cancer–specific survival [30]. In the current study, because many patients in this underinsured population do not have stable employment, current employment may indicate a more stable lifestyle and therefore the ability to comply with demands necessary to complete the recommended radiation therapy. Being a current cigarette smoker and larger pathologic size were again associated with poorer patient compliance with physician recommendations for radiation therapy. The effect of pathologic tumor size again likely reflects the effect of PC on that variable.

The strongest predictor of patient compliance with physician-recommended radiation therapy was patient compliance with chemotherapy (OR, 86.67; 95% CI, 9.21–814.7; p < .0001). The fact that compliance with chemotherapy was significantly associated with completion of radiation indicates that those who complete chemotherapy may be an intrinsically more compliant subpopulation.

Patients in the PC group were also more likely to initiate hormonal therapy than patients in the AC group (100% versus 62%; p = .0001). Univariate analysis of these matched patients revealed that being a current cigarette smoker was the only variable significantly associated with noncompliance with the initiation of hormonal therapy.

The better compliance with chemotherapy, radiation therapy, and hormonal therapy seen in patients who underwent PC has not been previously described. One reason may be that the majority of published breast cancer research involves non-Hispanic white, insured patients from major cancer centers. These patients are more apt to be compliant with physician recommendations, and similarly are the ones most likely to enroll in prospective clinical trials [19, 31]. Breast cancer patients, however, are heterogeneous with respect to both sociodemographic factors and the institutions where they receive their treatment. Therefore, in populations in which compliance with adjuvant treatments is suboptimal [18, 19, 21], any strategy to improve compliance should be considered. In this underinsured population, patients with a higher BMI and current cigarette smokers warrant further effort and better counseling to help facilitate compliance with physician-recommended adjuvant therapy. One possible explanation for the benefit of PC in terms of patient compliance may rest on a potential misconception among our patients, that surgery is curative and the end of treatment. Patients who complete PC may be more apt to grasp the concept that the operation is just one part of a multidisciplinary treatment approach to optimize their likelihood of survival.

This study has several limitations. Retrospective studies are subject to inherent limitations and biases. Data from prospective studies involving minority patients, however, are very difficult to obtain even in a disease as common as breast cancer. This study was a relatively small study at one institution with one patient population. Therefore, the applicability to other patient populations does need to be considered. Finally, follow-up was limited in this study and continued evaluation of patients is necessary to determine whether the better compliance with adjuvant therapy will translate into better long-term outcomes.

Conclusion

Breast cancer is a heterogeneous disease and patient populations are equally diverse. In these matched patients, PC led to better patient compliance with chemotherapy. In this underinsured population, patients who completed chemotherapy were also more likely to complete radiation therapy. Patients with a higher BMI and current cigarette smokers were more likely to have poor compliance with physician recommendations for adjuvant therapy.

Author Contributions

Conception/Design: Ian K. Komenaka, Chandra Jha, Boo Ghee Low

Provision of study material or patients: Ian K. Komenaka, Chandra Jha, Boo Ghee Low, Jason Salganick, Marcia Bouton

Collection and/or assembly of data: Ian K. Komenaka, Michael Hibbard, Chiu-Hsieh Hsu, Maria Elena Martinez, Marcia Bouton

Data analysis and interpretation: Chiu-Hsieh Hsu, Maria Elena Martinez, Jesse Nodora

Manuscript writing: Ian K. Komenaka, Michael Hibbard, Chiu-Hsieh Hsu, Maria Elena Martinez, Boo Ghee Low, Jason Salganick, Marcia Bouton, Jesse Nodora

Final approval of manuscript: Ian K. Komenaka, Michael Hibbard, Chandra Jha, Chiu-Hsieh Hsu, Maria Elena Martinez, Boo Ghee Low, Jason Salganick, Marcia Bouton, Jesse Nodora