Examination of a paradox: recurrent metastatic breast cancer incidence decline without improved distant disease survival: 1990–2011

We observed a sharp decline in rMBC incidence over time. Highest risk of distant relapse was associated with earlier time periods (1990–1998, 1999–2004), higher-stage disease at initial diagnosis (Stage II and III), hormone receptor negative or triple negative, younger age (< 40), high histologic grade tumors, or symptomatic disease at time of diagnosis. Disease characteristics at initial diagnosis (higher-stage, negative hormone receptor status, age 70 +) and severity of distant relapse (shorter DDFI, multiple sites, visceral distant disease) were significantly associated with worse distant disease survival. With decreased incidence, rMBC patient characteristics changed by tumor marker subtype (more triple negative, fewer hormone or HER2 positive) and rMBC disease profile (> 1 metastatic site, visceral disease, or combined local/regional/distant at relapse). Distant disease-specific survival did not improve over time.

In a cohort study, Wu et al conducted a multivariate analysis of any recurrence (local, regional, and distant) and observed declining recurrence risk over time associated with cancer detection method, tumor size, number of positive lymph nodes, and triple-negative tumor status [14]. In another cohort study, Cossetti et al observed declining all type recurrence over time, especially among HER2-positive and ER-negative/HER2-negative BC [15]. Chen et al identified stage III, progesterone receptor  negative, and HER2 negative as factors related to distant relapse at a median follow-up of 2.2 years and used their findings to develop a distant recurrence risk score [16]. Using any recurrence, two studies found young age (< 40) associated with higher hazard of recurrence [17, 18]. Other studies found hazard of distant recurrence nearly doubled for tumors found outside of screening (symptomatic disease) [19, 20].

Survival post distant recurrence is rarely examined as few studies have prospective data on outcomes other than singular mortality. In an extensive study of patients treated in adjuvant chemotherapy trials and followed for outcomes (ECOG) (1978–2002), no improvement in survival post distant recurrence was observed over time [7]. Giordano, using institutional cohort data (1974–2000), and Gennari, using clinical trial data (1983–2001), both found suggestive but non-significant survival improvement over time post distant recurrence treatment [5, 6]. In our Cox proportional hazards model, diagnosis year was not associated with post-rMBC diagnosis survival and no survival improvement post-rMBC diagnosis was observed. The link between triple-negative disease and visceral metastasis has been well documented although we found both factors to be independently related to shorter distant disease-specific survival [21].

With the advent of dose-dense adjuvant chemotherapy regimens, taxanes, and effective targeted therapy for both hormone receptor-positive and HER2-positive disease in the adjuvant/neoadjuvant setting, rMBC characteristics have changed coincident with the observed incidence decline [22, 23]. The observed reduction in both HER2 + and HR + rMBC incidence over time indicates increased success of initial targeted therapy with trastuzumab and hormone therapy [24, 25]. Our observation of decreased HER2 + rMBC incidence in the most recent time period is consistent with reported results of improved long-term outcomes after neoadjuvant/adjuvant treatment with HER2-targeted therapy [26]. Improvements in early disease treatment targeted at hormone receptor and HER2-positive disease have reduced overall distant relapse rates but resulted in a new profile of rMBC disease that may be more aggressive with a higher chemoresistant probability [27, 28].

The primary strength of our study is the level of patient and treatment detail with long-term follow-up of all cases for distant recurrence and vital status carried out by a dedicated breast cancer registry staff. A breast cancer registry cohort with continuous follow-up over a 26-year time period provides the opportunity to evaluate changes in incidence and outcomes over a period in which breast cancer treatment has become more effective and personalized. The use of both 5- and 10-year rates of distant disease incidence captures both early (hormone receptor negative) and late (hormone receptor positive) distant disease recurrence, with our DRFS KM plots extending to 15 years of follow-up. The use of hazard modeling allows for simultaneous evaluation of factors presumed to affect the outcomes of interest, distant recurrence and survival time post distant recurrence. Diagnosis year intervals, aligned with timing of known significant treatment changes, were used as factors for therapy advances in the statistical analysis.

Colleoni et al found a lower annualized hazard of recurrence among ER-positive versus ER-negative breast cancer in the first 5 years post-diagnosis which reversed at 10 years post-diagnosis [29]. With less follow-up, our last cohort (2005–2011) findings may be skewed towards more triple-negative and less HR+ distant recurrence due to differential time to distant relapse associated with hormone receptor status. For this reason, we ran additional analysis using only the first two cohorts and Kaplan–Meier plots with at least 10-year follow-up and found no difference in proportional change over time. We do not know if mammography or symptomatic patients participated in regular breast cancer-screening programs. Our observations from an NCCN guideline compliant institution treating a high SES population with less diversity than other metropolitan areas may contribute to better outcomes than those achieved in other parts of the US making our results less generalizable [30, 31].

The observed fifty percent relative decline in distant breast cancer recurrence over time may be related to both improved treatment for initial disease at diagnosis decreasing recurrence risk (hormonal therapy, polychemotherapy, dose-dense chemotherapy, taxanes, and trastuzumab) and stage shift to more early and less late stage disease at diagnosis with improved screening technology and screening program participation [32]. Distant recurrent disease incidence decline over time differed by phenotypic characteristics. We observed a 70% decline in distant disease recurrence among hormone receptor-positive patients and a 50% decline among hormone receptor-negative patients over time. We observed a 60% decline in distant disease recurrence among HER2 positive patients from 1999 to 2006. The differential decline associated with phenotypic subtype creates a new profile of recurrent metastatic breast cancer with fewer HR and HER2-positive cases and relatively more TNBC cases. Ten-year cumulative incidence comparisons to accommodate the longer interval to distant recurrence among HR-positive versus HR-negative disease did not significantly differ from 5-year rates.

The observed change in the current rMBC population to more triple-negative disease, simultaneous local/regional/distant recurrent disease, 2 or more metastatic sites at distant recurrence and visceral disease presents a challenging clinical situation. A more severe metastatic disease profile could be due to better imaging, more TNBC, and distant recurrence among chemoresistant cases. Distant disease survival worsened over time, significant at the .03 level. If high-level guideline compliant care can be universally delivered, on balance rMBC incidence would be expected to decline nationally especially in regards to HR- and HER2-positive appropriate treatment but it is not expected a similar improvement in distant disease survival will be observed.

The development of CDK 4/6 inhibitors for HR+ metastatic disease would be expected to result in longer DDFS for HR+ patients. Their inclusion in neoadjuvant and adjuvant therapy regimens may also reduce the incidence of rMBC [33]. New treatment developments hold promise for extension of life post distant disease diagnosis but the increased challenge of a poor prognosis pool of rMBC cases is a serious problem [34‐38]. Clinical challenges include treatment of rMBC patients that are older, previously treated, have multiple metastatic sites and/or visceral metastases at time of distant recurrence. Separate analysis of rMBC and de novo Stage IV MBC in clinical trial evaluation is warranted as these two MBC subtypes may have differential response to treatment [4].

Breast cancer that is higher-stage, symptomatically detected, high histologic grade, triple-negative, or young age at initial diagnosis is at greatest risk for distant relapse. Significant progress has been made in the diagnosis and treatment of breast cancer with guideline compliant treatment as measured by our observation of declining distant recurrence incidence over time. These findings support both screening for early detection and improved treatment over the last three decades. The lack of progress in extending survival for recurrent metastatic breast cancer is disappointing but likely due to the worsening profile of recurrent metastatic disease with tumor chemo-resistance and absence of viable treatment options related to patient or metastatic tumor characteristics. Development of targeted therapies for initial triple-negative disease, screening and breast awareness, treatment guideline compliance, and identification of recurrent metastatic disease treatment are all important goals. At present, guideline adherence and appropriate targeted therapy hold the most potential for distant recurrence reduction and subsequent breast cancer mortality improvement.