Low-Intensity Adjuvant Chemotherapy for Breast Cancer in Older Women: Results From the Prospective Multicenter HOPE Trial

PURPOSE

Older women with high-risk early breast cancer (EBC) benefit from adjuvant chemotherapy, but their treatment is frequently complicated by toxic side effects, resulting in dose reductions and delays. This makes it challenging for oncologists to maintain a relative dose intensity (RDI) ≥ 85%, as recommended for optimal curative-intent treatment. Understanding which women are at risk of receiving suboptimal RDI may inform treatment discussions and guide early, targeted supportive care or geriatric comanagement interventions.

METHODS

This was a prespecified secondary analysis of the HOPE trial, which enrolled women age ≥ 65 years with EBC initiating neoadjuvant or adjuvant chemotherapy. RDI was calculated as the ratio of delivered to planned chemotherapy dose intensity. The primary outcome was low RDI, defined as RDI < 85%. Multivariable logistic regression with stepwise selection was used to evaluate the association between baseline variables (demographic, clinical, and geriatric assessment) and low RDI. Survival probability was estimated using the Kaplan-Meier method, and the log-rank test was used to compare overall survival.

RESULTS

Three hundred twenty-two patients (median age at diagnosis, 70 years; range, 65-86 years) were included. The median follow-up was 4 years. Sixty-six patients (21%) had a low RDI. Age ≥ 76 years (odds ratio [OR], 2.57; 95% CI, 1.12 to 5.91; P = .03), lower performance status (OR, 4.32; 95% CI, 1.98 to 9.42; P < .001), and use of anthracycline-based or cyclophosphamide, methotrexate, and fluorouracil regimens (OR, 3.47; 95% CI, 1.71 to 7.05; P < .001) were associated with low RDI. The 5-year overall survival probability was 0.80 versus 0.91 in patients with RDI < 85 versus ≥ 85%, respectively (log-rank P = .02).

CONCLUSION

One in five older patients with EBC treated with standard chemotherapy received low RDI and had inferior survival outcomes. Older patients at risk for low RDI should be identified and targeted upfront before initiating chemotherapy.

INTRODUCTION

Women with high-risk early breast cancer (EBC) benefit from adjuvant chemotherapy, irrespective of age.^1,2 However, compared with younger and healthier women, older and more frail women with EBC experience increased chemotoxicity, resulting in dose reductions and treatment delays.^3-5

CONTEXT

• Key Objective

• Women with high-risk early breast cancer (EBC) benefit from adjuvant chemotherapy, irrespective of age. However, compared with younger women, older women with EBC experience increased chemotoxicity and often receive lower than intended relative dose intensity (low RDI) of chemotherapy. Our key objective was to evaluate the incidence of low RDI, the risk factors associated with low RDI, and the relationship between low RDI and survival outcomes in older adults with EBC.

• Knowledge Generated

• In this secondary analysis of a prospective cohort, one in five older women with EBC treated with standard chemotherapy received low RDI and had inferior overall survival.

• Relevance (K.D. Miller)

• Clinicians often have a lower threshold to reduce chemotherapy dose intensity in older patients to reduce the risk of serious toxicity, but lower doses may reduce the efficacy of treatment. This study identifies patients most likely to require dose reductions, so patients who chose to proceed with chemotherapy can receive additional supportive care proactively.*

  *Relevance section written by JCO Senior Deputy Editor Kathy D. Miller, MD.

Patients with EBC who require dose reductions and treatment delays are more likely to experience disease recurrence and death than patients who complete their full chemotherapy dose on time.⁶ Relative dose intensity (RDI) is a metric that combines the total dose administered with adherence to the planned treatment schedule. Compared with women receiving ≥ 85% of RDI, women with EBC receiving < 85% of RDI have a 57% increase in the risk of disease recurrence at 10 years.^7-9

Chronological age is a well-established risk factor for low (< 85%) RDI.¹⁰ However, the incidence, risk factors, and outcomes of low RDI in older women with EBC have been inconsistent and are limited to retrospective analyses. EBC trials showed that women who received a low RDI had inferior survival outcomes.^6,7,11-16 But these trials rarely included older adults. Indeed, most data on RDI in older patients with breast cancer are retrospective.^17-21 The largest of these studies demonstrated that 75% of patients age ≥ 65 years received low RDI.¹⁹ By contrast, other retrospective analyses showed that older patients with EBC tolerate newer-generation adjuvant regimens and many maintain RDI ≥ 85%.^18,21-25

To advance existing evidence, we leveraged a multicenter prospective study of older women (age ≥ 65 years) with EBC treated with current standard neoadjuvant/adjuvant chemotherapy to evaluate the (1) incidence of low RDI, (2) risk factors associated with low RDI, and (3) relationship between low RDI and overall survival (OS) as well as breast cancer–specific (BCS) and non–breast cancer-related mortality.

METHODS

The Hurria Older PatiEnts with Breast Cancer (HOPE) Study (ClinicalTrials.gov identifier: NCT01472094) is a prospective multicenter study of patients age 65 years and older with EBC treated with current standard neoadjuvant/adjuvant chemotherapy regimens, which rigorously captured detailed geriatric, clinical, and treatment data. This Institutional Review Board–approved study was conducted by the Cancer and Aging Research Group (CARG) at 16 sites across the United States and has been reported previously.^5,26 Here, we examine the incidence, risk factors, and survival outcomes associated with low RDI in participants enrolled in the HOPE cohort. This was a prespecified secondary analysis.

Study Population

Participants (N = 501) age ≥ 65 years with stage I-III breast cancer and starting neoadjuvant or adjuvant chemotherapy were recruited between September 2011 and May 2017. For this analysis, we excluded patients with human epidermal growth factor receptor 2–positive (HER2-positive) disease (n = 121) and HER2-equivocal disease treated with trastuzumab (n = 7) because of the potential differences in the effect of RDI on survival between patients with HER2-positive and HER2-negative tumors. In addition, those treated with nonstandard regimens, defined as regimens outside of the National Comprehensive Cancer Network guidelines (n = 28), those who had upfront dose reduction (n = 18), and those who changed treatment plans after initiating therapy (n = 4) were excluded. The final analytic sample composed of 322 patients (Data Supplement, online only).

Prechemotherapy Measures

Before the start of neoadjuvant/adjuvant chemotherapy, we collected demographic and clinical variables, including age, race/ethnicity, education, marital status, weight, height, body mass index, body surface area (BSA), cancer stage, estrogen/progesterone/HER2 receptor status, WBC count, hemoglobin, albumin, creatinine clearance, blood urea nitrogen, liver function (normal, defined as all liver function tests within normal ranges for each institution), and use of primary granulocyte-colony stimulating factor (G-CSF) prophylaxis (yes/no).

We used the CARG geriatric assessment (GA)^27,28 to collect baseline objective and self-reported measures of functional aging (Data Supplement).

Outcomes

Patients were followed throughout chemotherapy, from start to 30 days postcompletion of chemotherapy. The treating oncologist prescribed the chemotherapy regimen. Before starting chemotherapy, the planned regimen, dose, and cycles were documented. Throughout treatment, the actual regimen received, including chemotherapy drugs and doses, start and end dates of each cycle, and the number of cycles received were captured.

The primary end point was the proportion of patients receiving suboptimal or low RDI, defined as RDI < 85%. On the basis of previous studies, RDI was calculated as the ratio of actual chemotherapy dose intensity received to the standard dose intensity, ranging from 0% to 100%.^19,22 Chemotherapy dose intensity was defined as the amount of drug (mg) per unit of BSA (m²) per unit of time (week).^12,29 To calculate the standard dose intensity (mg/m²/week), the total standard chemotherapy dose for each drug was divided by the standard duration of that drug, including all planned cycles as recommended by National Comprehensive Cancer Network guidelines. To calculate actual dose intensity (mg/m²/week), the total received chemotherapy dose (accounting for dose modifications) for each drug was divided by the actual duration (accounting for dose delays) for which each patient received that treatment. The RDI for each patient represents the average RDI of all chemotherapeutic agents in the received regimen. Given that we excluded all patients who had upfront, planned dose reductions, the difference between the planned (standard) dose intensity and the actual dose intensity delivered defines subsequent, unplanned reductions in dose intensity.

The secondary end points for this study were OS, BCS mortality, and non–breast cancer-related mortality. Survival time was calculated from the start of chemotherapy to the date of death because of any cause, the date of last contact, or the end of follow-up date (December 31, 2019), whichever occurred first.

Statistical Analysis

The primary outcome of this study was low RDI. RDI was dichotomized using 85% as the cutoff point, and < 85% was defined as low RDI.⁷ Incidence was defined as the proportion of patients with low RDI, assessed across all regimens and participants and by the specific treatment regimen.

To evaluate risk factors associated with low RDI, we first conducted a univariate logistic regression to examine the relationship between baseline variables (demographic, clinical, and GA) and low RDI. Age was examined as both a continuous and dichotomized variable. Youden's index was used to determine the best cutoff point (75 years) with the highest sensitivity and specificity for age in relation to low RDI. Multicollinearity was tested using the CORRB option from the logistic regression model among variables with P < .05.^30,31 Baseline variables were not highly correlated, except for the planned treatment duration and regimen. Given that these two variables (planned treatment duration and regimen type) were highly correlated, only the regimen type was included in the stepwise selection. Variables that satisfied P < .05 criteria in univariate analysis were included in multivariable logistic regression analysis with a stepwise selection (backward and forward). To minimize overfitting, we chose to take a conservative approach using a more stringent P value cutoff (.01) for the final significance³²; variables with P < .01 in the stepwise selection model were retained in the final model. On the basis of our previous research,⁵ the final model included clinically meaningful variables, such as age, stage, liver and kidney function, history of cardiovascular disease, and stepwise selected variables. Other nonstatistically significant covariates were not retained.

Finally, we evaluated associations between low RDI and secondary end points (OS, BCS morality, and non–breast cancer-related mortality). OS probability was estimated using the Kaplan-Meier method, and the log-rank test was used to compare the survival curves for low RDI versus RDI ≥ 85%. Univariate and multivariable Cox regressions adjusting for breast cancer subtype were used to examine the association between low RDI and OS. The relationship between low RDI and cause-specific mortality was examined using the cumulative incidence function, where breast cancer and non–breast cancer causes were considered as competing events for each other. Hazard ratios for low RDI were calculated using the Fine and Gray regression models,³³ allowing for a subdistribution of hazards of death because of breast cancer when considering death because of other causes as a competing event and vice versa. P values < .01 were considered statistically significant. All analyses were conducted using SAS Version 9.4 (SAS Institute Inc, Cary, NC).

RESULTS

Participant Characteristics

Participants ranged from age 65 to 86 (median 70) years and had an average of two comorbidities (range, 0-11; Table 1). Most were non-Hispanic White (74%) and had stage II (44%) or III (22%) disease. Most participants (83%) were treated with chemotherapy in the adjuvant setting, with 47% receiving docetaxel + cyclophosphamide (TC),^34,35 47% receiving anthracycline-based regimens (eg, doxorubicin + cyclophosphamide [AC] followed by paclitaxel [T] or AC alone),^36-38 and 6% receiving cyclophosphamide, methotrexate, and fluorouracil (CMF)^36,39 (Fig 1). There was no difference in TC versus AC with or without T use by breast cancer subtype. Most participants (85%) received primary G-CSF prophylaxis with the first cycle of chemotherapy.

TABLE 1.

Baseline Characteristics of Older Women Receiving Chemotherapy for Early Breast Cancer and RDI of Chemotherapy Delivered

FIG 1.

Distribution of chemotherapy regimens administered among 322 patients age 65 years and older with HER2-negative early breast cancer. CMF once every 3 weeks = IV cyclophosphamide, methotrexate, and F (n = 4) and CMF once every 2 weeks = IV cyclophosphamide, methotrexate, and F (n = 16). One patient received oral cyclophosphamide and IV methotrexate + F on a 28-day cycle. A, doxorubicin; C, cyclophosphamide; dd, dose dense; F, fluorouracil; HER2, human epidermal growth factor receptor 2; IV, intravenously; M, methotrexate; other anthracycline-based regimens include epirubicin and cyclophosphamide (n = 1), and fluorouracil, doxorubicin, and cyclophosphamide (n = 3); T, paclitaxel; TC, docetaxel + cyclophosphamide.

Incidence of Low RDI

Of the 322 patients who began chemotherapy at the standard dose, 66 (21%) experienced unplanned reductions in RDI and ultimately received < 85% of the standard dose intensity. The mean and median RDI administered to all patients were 90% and 100%, respectively (quartile 88-100, range, 18-100). More than half (52%) of the patients received a RDI of 100%. Hormone receptor-positive/HER2-negative and TNBC had similar RDI (P = .34). Figure 2A shows the distribution of the RDI across all regimens and participants.

FIG 2.

Distribution of RDI (A) overall (total cohort [n = 322]) and by regimen: (B) TC (n = 151), (C) anthracycline-based regimens (n = 150), and (D) CMF (n = 21). CMF, cyclophosphamide, methotrexate, and fluorouracil; RDI, relative dose intensity; TC, docetaxel + cyclophosphamide.

Figures 2B-2D show the distribution of RDI by chemotherapy regimen. Compared with patients who received TC, patients who received anthracycline-based and CMF regimens were more likely to have a low RDI (13% TC v 28% anthracyclines v 19% CMF, P < .01). Similarly, compared with patients with the planned treatment duration of < 12 weeks, those with planned treatment durations of ≥ 12 weeks were more likely to have low RDI (14% v 27%, P = .004).

Dose reduction, treatment delay, and early discontinuation were the actions that led to low RDI. Among patients with any dose reduction or any treatment delay, 42.4% and 44.3% had low RDI, respectively. Among patients with early discontinuation, 75.8% had low RDI. Among patients with treatment delay alone, 16.7% had low RDI, whereas among those with dose reduction alone, 11.5% had low RDI (Data Supplement).

Risk Factors Associated With Low RDI

In bivariate analysis, low RDI was associated with older age (≥ 76 years), higher stage (II/III), use of non-TC regimens (anthracycline-based or CMF), planned treatment duration > 12 weeks, abnormal baseline liver function (liver function tests outside reference normal ranges), low kidney function (glomerular filtration rate, < 60), worse physician-rated Karnofsky performance status (KPS; < 90), poor physical function (ADL score < 70), and a history of cardiovascular disease (Table 2). Baseline BSA, WBC, use of primary G-CSF prophylaxis, hormone receptor status, prior chemotherapy, and prior radiation were not associated with low RDI. Although not statistically significant, patients who received primary G-CSF prophylaxis were slightly less likely to have low RDI (19% v 28.6%, P = .13).

TABLE 2.

Univariate and Multivariable Analyses of Baseline Characteristics in Relation to Low RDI Among Older Women Early Breast Cancer

Multivariable stepwise logistic regression analysis identified that age ≥ 76 years, use of anthracycline-based or CMF regimens, and physician-rated KPS < 90 were associated with low RDI at P < .01 (Table 2). Compared with patients age 75 years or younger, the odds of low RDI were three times greater for those age 76 years or older (odds ratio [OR], 3.15; 95% CI, 1.41 to 7.06; P = .005). Compared with those with KPS 90-100, the odds of receiving low RDI was 5.8 for those having a physician-rated KPS < 90 pretreatment (OR, 5.80; 95% CI, 2.77 to 12.17; P < .001). Finally, compared with the TC regimen, the odds of receiving low RDI was four times greater for those receiving anthracycline-based or CMF regimens (OR, 3.90; 95% CI, 1.98 to 7.67; P < .001).

In the final model, after accounting for clinically meaningful variables such as stage, liver function, kidney function, and history of cardiovascular disease, the results for physician-rated KPS (OR, 4.32; P < .001) and use of anthracycline-based or CMF regimens were still highly significant (OR, 3.47; P = .001), whereas the association for older age was slightly lessened (OR, 2.57; P = .03) (Table 3).

TABLE 3.

Final Multivariable Model for Low Relative Dose Intensity in Older Women With Early Breast Cancer

Low RDI and Survival

By the end of 2019 with a median follow-up time of 4.1 years (range, 63 days-8.0 years), 31 patients were deceased. Sixteen patients died from breast cancer and 15 from non–breast cancer causes. Compared with patients with RDI ≥ 85%, patients with low RDI had a higher risk of both breast cancer–specific mortality (hazard ratio [HR], 2.18; 95% CI, 0.77 to 6.17; P = .14) and non–breast cancer-specific mortality (HR, 2.37; 95% CI, 0.83 to 6.78; P = .11), but neither reached statistical significance (Data Supplement).

Compared with patients with RDI ≥ 85%, those with low RDI had significantly lower OS (HR, 2.36; 95% CI, 1.14 to 4.87; P = .02; Fig 3). In a multivariable Cox regression model, after adjusting for the breast cancer subtype, low RDI was still significantly associated with worse OS (HR, 2.47; 95% CI, 1.20 to 5.10; P = .01). The 2-year and 5-year survival probabilities were 0.91 and 0.80 in patients with low RDI compared with 0.99 and 0.91 in patients with RDI ≥ 85%. A similar association between low RDI and OS was observed after excluding patients age 76 or older (Data Supplement, log-rank P = .02); the 2-year and 5-year survival probabilities were 0.91 and 0.79 among patients with low RDI compared with 0.99 and 0.92 in patients with RDI ≥ 85%.

FIG 3.

Estimated survival probability by RDI. RDI, relative dose intensity.

DISCUSSION

In this prospective cohort of more than 300 patients age 65 years and older with EBC treated with standard-of-care neoadjuvant/adjuvant chemotherapy, one in five patients received low RDI. Factors associated with low RDI were age, performance status, and treatment regimens, specifically use of anthracyclines or CMF. Older women with low RDI had higher mortality than women who received at least 85% RDI. However, there were not enough cause-specific events to determine whether this increased mortality was due to breast cancer or other causes.

These findings have several important clinical implications. First, this study demonstrates that most, but not all, older patients with EBC who receive chemotherapy maintain at least 85% RDI. This finding differs from previous studies, which showed that up to 65% of older patients could not maintain optimal RDI.^17-19,21 However, most of these studies were retrospective, included patients with upfront, planned reduced RDI, and largely examined first-generation EBC regimens not commonly used today. By contrast, most patients in our cohort were treated with current standard chemotherapy regimens. Moreover, data on RDI with taxane-based adjuvant chemotherapy regimens in the older adult population remain very limited. In our analysis, 47% of patients received TC, adding important information to the literature.⁴⁰ Here, we show that nearly 80% of patients with breast cancer age ≥ 65 years received an RDI ≥ 85% with current standard chemotherapy regimens. Our findings provide encouraging evidence that maintaining an optimal RDI is achievable in older patients. However, few women in this sample were above age 75 years and most were relatively fit at baseline.

Second, not all older women achieve optimal RDI. Despite advances in supportive care, GCSF prophylaxis (85% of participants received primary G-CSF prophylaxis), and improved antiemetics, approximately 20% of older women still receive < 85% RDI and have inferior outcomes. These women are often left with short- and long-term adverse effects of chemotherapy and little survival benefit. Identifying these women upfront may facilitate personalization of cancer treatment and inform shared decision making.

In addition, models that predict severe chemotherapy toxicity, such as the CARG and CARG-BC chemotoxicity risk scores, may identify potentially modifiable determinants of chemotoxicity, including physical and cognitive function, social support, and falls.^5,28,41 The CARG and CARG-BC scores are built on models derived from the GA and are highly predictive of low RDI.⁵ Although the purpose of this study was to identify individual variables from the GA associated with low RDI rather than composite scoring systems, we did find significant associations between CARG and CARG-BC scores and low RDI (Data Supplement). By integrating this patient health–related information into treatment decisions, we may be able to optimize chemotherapy delivery and survivorship-related outcomes.

Third, our findings highlight the need for targeted interventions to preserve dose intensity among older patients with breast cancer. To date, most of the work around cancer treatment toxicity in the geriatric oncology population has been observational—centered on the use of the GA to quantify risk of chemotoxicity,^5,41 evaluate barriers to trial enrollment,^42-44 and assess health status in older patients with cancer.^45-47 The role of the GA as a predictor of adverse clinical outcomes is well established, and recent studies have shown that GA-guided interventions reduce chemotoxicity.^48,49 However, the uptake of the GA remains low, and barriers to integrating GA in clinical care are prevalent.^40,50,51 Moreover, efforts to develop mechanism-based interventions to prevent, delay, or reverse the adverse consequences of cancer treatment remain limited.⁵² Targeted interventions aimed at minimizing chemotoxicity and optimizing RDI in older women with EBC are needed.

There are several limitations to this study. First, although there were no limitations on enrollment on the basis of performance status, most patients who participated had a physician-rated KPS ≥ 90, likely because of being preselected by their oncologists to receive chemotherapy. Second, the 16 sites involved in this study were tertiary care centers. Thus, the results may not be generalizable for patients in community sites. Third, our study sample size is relatively small, limiting our ability to examine a wider range of variables that could be associated with low RDI. In particular, abnormal liver and kidney function, routinely evaluated factors expected to predispose patients to low RDI, were ultimately not statistically significant despite an absolute difference between the two groups. Fourth, with a short follow-up time and limited death events, our analysis of RDI and survival is limited. The observed nonsignificant increase in the hazard ratio for both BCS and non–breast cancer-related mortality among patients with low RDI cannot be used to draw practice-changing conclusions. We realize the importance of this distinction as it answers the question of whether patients had inferior survival because they did not derive the full benefit of therapy or if the toxicity associated with low RDI led to persistent decline and early mortality. To address this question, we plan to extend our follow-up and collect long-term OS and cause-specific survival data. Finally, we selected 85% as the cutoff point for this study on the basis of previous studies, but more work is needed to clarify if the RDI threshold of 85% is appropriate for use in the older adult population.

In conclusion, one in five older patients with EBC treated with standard chemotherapy received low RDI and had inferior survival outcomes. This study informs clinical practice by highlighting the importance of preserving dose intensity in older patients with breast cancer. Moreover, older patients at risk of receiving low RDI may be identified and targeted upfront before initiating chemotherapy. Understanding safe and optimal strategies to minimize toxicity without sacrificing efficacy should be a priority for all patients.

AUTHOR CONTRIBUTIONS

Conception and design: Mina S. Sedrak, Harvey J. Cohen, Cary P. Gross, William P. Tew, Beverly Moy, Nicole M. Kuderer, Gary H. Lyman, Allison Magnuson Hyman B. Muss

Financial support: Mina S. Sedrak, William Dale

Administrative support: Mina S. Sedrak, William Dale, Vani Katheria

Provision of study materials or patients: Mina S. Sedrak, Heidi D. Klepin, Rachel A. Freedman, Tracey O'Connor, Mary Ann Fenton, Beverly Moy, Allison Magnuson Hyman B. Muss

Collection and assembly of data: Mina S. Sedrak, Jingran Ji, William P. Tew, Heidi D. Klepin, Tanya M. Wildes, Efrat Dotan, Rachel A. Freedman, Tracey O'Connor, Mary Ann Fenton, Beverly Moy, William Dale, Vani Katheria, Allison Magnuson Hyman B. Muss

Data analysis and interpretation: Mina S. Sedrak, Can-Lan Sun, Jingran Ji, Harvey J. Cohen, William P. Tew, Tanya M. Wildes, Efrat Dotan, Rachel A. Freedman, Tracey O'Connor, Selina Chow, Beverly Moy, Andrew E. Chapman, William Dale, Nicole M. Kuderer, Gary H. Lyman, Allison Magnuson Hyman B. Muss

Manuscript writing: All authors

Final approval of manuscript: All authors

Accountable for all aspects of the work: All authors

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

Low-Intensity Adjuvant Chemotherapy for Breast Cancer in Older Women: Results From the Prospective Multicenter HOPE Trial

The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated unless otherwise noted. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/jco/authors/author-center.

Open Payments is a public database containing information reported by companies about payments made to US-licensed physicians (Open Payments).