Introduction

As the prevalence of women living with HIV rises, increase attention is needed to the long-term health maintenance and care in this specific population. In the general population, breast cancer is the most common malignancy in women (excluding skin cancer) in the US. and the second most common cause of cancer-related death in women with an estimate of 232,670 new cases to be diagnosed in 2014, causing about 40,000 deaths each year from the disease [1–3]. It is well established that since the introduction of highly active antiretroviral therapy (HAART), the life expectancy of HIV-infected individuals has increased, while the incidence of both opportunistic infections and AIDS-defining cancers (ADCs) has dramatically decreased [4, 5] With this shift, the rate of Non AIDS-Defining Cancers (NADCs) has risen and now accounts for the majority of cancers in HIV-infected persons [6]. For breast cancer, however, initial reports documented a decrease incidence in cohorts from HIV patients relative to the general population [7–10]. Later studies have suggested that the risk has increased significantly in the last years and is now approaching the risk of the general population [11–16]. According to the latest report of the World Health Organization [17], the proportion of women living with HIV has been increasing in the last 10 years and now constitutes 50 % of people living with HIV worldwide, the cases of breast cancer in this population are therefore expected to increase.

Treating patients with HIV and breast cancer remain a challenge. In 2001, we reported our experience with 20 patients with breast cancer and HIV and it was the largest cohort to that date [18]. Since then, even though there have been continuous reports, it still remains the report of small case series and a few series linking cancer to HIV registries [9, 11, 13, 15, 19–25]. The recent report of 151 patients in Soweto, Africa [12] constitutes the largest cohort to date and they reported equal prevalence to the HIV-negative cohort; however, no information regarding treatment and outcome was given. There are currently no approved guidelines or recommendations for screening or for treating HIV patients with breast cancer and clinicians must make decisions despite the lack of well-validated data regarding outcomes, tolerability, safety, interactions, and management of predictive clinical variables unique to HIV patients like CD4 and viral load [11, 26].

We therefore aim to present an updated report of our experience with 48 cases of HIV-seropositive women with carcinoma of the breast diagnosed between January 1, 1989 and December 31, 2013 in order to share the key points learned regarding presentation, choice of therapy, and expected complications and outcomes.

Patients and methods

Study population and study design

This study was approved by the University of Miami IRB. This was a retrospective registry-based analysis that included all HIV-positive patients who received their breast cancer care at the University of Miami/Jackson Memorial Hospital (UM/JMH) between January 1, 1989 and December 31, 2013. Cases were identified by a review of the JMH Oncology Clinic patient roster and the breast cancer clinic patient registry from Sylvester Comprehensive Cancer Center (SCCC). All cases were confirmed by chart review. HIV testing was not routinely performed. HIV infection was determined by enzyme-linked immunosorbent assay (ELISA), with confirmation by Western blot analysis. Breast cancer diagnoses were confirmed by histology. Cases in which HIV status had been diagnosed after initial treatment were excluded from the analysis.

Definitions and endpoints

Data were collected on patient demographics, HIV, disease and treatment characteristics, and clinical outcome. Demographic variables included age at diagnosis and race/ethnicity. HIV-related variables included date of diagnosis, CD4+ count, and viral load (VL) at breast cancer diagnosis, risk factors for HIV (Intravenous drug abuse, unprotected sexual intercourse, and Blood transfusion) and history of any AIDS-defining condition (ADC). AIDS was defined by CDC case definition as a CD4 count < 200 or any history of ADC. HAART therapy at diagnosis was established.

Cancer-related variables included T, N, M, stage, Estrogen (ER), and Progesterone (PR) receptor expression evaluated by immunohistochemistry and Her-2/neu expression evaluated by either fluorescence in situ hybridization (FISH), Chromogenic in situ hybridization (CISH) or IHC. Type of chemotherapy, targeted therapy (either hormonal or Her-2 directed), and radiation was recorded for each patient. Progression-free survival (PFS) was defined as the time from cancer diagnosis to relapse, disease progression, or death from any cause. Overall survival (OS) was defined as the time from cancer diagnosis to death from any cause. Death was confirmed by clinical records, the Social Security Death Index, or family member, and cause of death was classified as related to breast cancer, AIDS, other clinical conditions and unknown. Breast cancer-specific survival was defined as the time from diagnosis to the time of a breast cancer-related death.

Statistical analysis

Patient, HIV- and cancer-related variables were described as categorical, discrete, or continuous variables as appropriate with corresponding measurements of central tendencies, absolute and relative frequencies, and dispersion measures (Interquartile range (ICR) for non-normal data and standard deviation (SD) for normal data). Discrete estimates were rounded to the closest whole number. We calculated 95 % confidence intervals (CI) or standard error (SE) where appropriate. The probabilities of PFS, OS, and Breast cancer-specific survival were estimated from the time of Breast cancer diagnosis using Kaplan–Meier survival curves and calculating their confidence interval. Univariate analyses of the effect of stage on PFS, OS, and breast cancer-specific survival were analyzed by Cox proportional hazard ratios (HR); Multivariable analysis was not pursued due to the small cohort. Data were analyzed by complete case analysis and no imputation data were used for missing data. Stata 12.0 was used for statistical analysis.

Results

A total of 55 possible cases were identified by review of the JMH Oncology Clinic patient roster and the breast cancer clinic patient registry from SCCC between January 1, 1988 and December 31, 2013, and were included for chart review. Of these, 2 were excluded because they were HIV-negative patients, 3 because HIV was diagnosed after treatment, and 2 because chart was unavailable for review. A total of 48 cases were included in the analysis.

Baseline patient characteristics

Demographic, HIV- and cancer-related variables are summarized in Table 1.

Table 1 Demographic, HIV- and cancer-related variables with staging and prognostic information at baseline
Full size table

Demographics

There were 47 female patients and 1 male patient with breast cancer and HIV infection. The median age of breast cancer diagnosis was 46 years (IQR 41–52). Thirty females (64 %) were premenopausal. 65 % were African American, 19 % Hispanic, 10 % Caribbean, and 6 % were Caucasian.

HIV related

The initial date of diagnosis of HIV infection was available for 30 patients (63 %). Median time of HIV diagnosis to Breast Cancer diagnosis was 4 years (ICR 1–11 years). A CD4+ count in patients with HIV/AIDS at the time of initial breast cancer diagnosis was available in 39 patients. Median CD4+ count was 330 cells/µL (IQR 131–589 cells/µL). A viral load was available in 16 patients at time of diagnosis: an undetectable viral load in 6 patients and in the other 10 ranged between 949 and 299,000 copies/µL. 20 patients (41 %) had documented AIDS at time of diagnosis of breast cancer; 15 with CD4 count below 200 cells/µL ± ADC and 5 additional had a CD4+ count > 200 cells/µL with a documented history of ADC (Pneumocystis jirovecii pneumonia in 2, Cytomegalovirus retinitis, Toxoplasmosis of brain, and esophageal candidiasis in one patient each). Risk factors for HIV were not reported in half of the charts reviewed and HAART was only reported in 20 patients at diagnosis. Of these, only 12 were on active HAART treatment on diagnosis.

Cancer related

Two percent of patient had Stage 0 disease, 8 % Stage I, 44 % Stage II, 40 % Stage III, and 6 % had stage IV. The median tumor size was 3.6 cm (IQR 2.8–6.7 cm), and clinical or pathological lymph node involvement was presented in 31 patients (65 %).

Treatment

Discrimination of treatment by stage is shown in Table 2.

Table 2 Treatment discriminated by stage
Full size table

Locoregional disease (n = 45)

For the 45 patients who presented with non-metastatic disease, 14 underwent neoadjuvant chemotherapy (31 %) and one patient had progressive disease during neoadjuvant treatment. Definitive surgery was recommended for 44 patients but 3 patients declined. In the other 41 patients who consented to surgery, modified radical mastectomy or breast conservative therapy (BCT) was performed in 31 and 9 patients, respectively. One patient opted for bilateral mastectomy.

Adjuvant treatment included chemotherapy in 18 (40 %), hormonal therapy alone in 8 patients (18 %) and in combination with chemotherapy in 13 (29 %), Her-2 directed in 1 with chemotherapy and radiation (XRT) in 14 patients (32 %). Seven patients refused either chemotherapy or radiation. Of the 29 patients with stages I–III disease, who had received chemotherapy as neoadjuvant or adjuvant, 24 received an anthracycline-based chemotherapy.

Late stage breast cancer (n = 3)

Three patients presented with stage IV disease at the time of initial breast cancer diagnosis and all were ER+. 1 patient received anthracycline-based chemotherapy and died from neutropenic fever during the first cycle, the other 2 received endocrine therapy alone.

Hormonal and Her-2 over expression

ER, PR, and Her-2 status were available for 44, 35, and 27 patients, respectively. 25 tumors (52 %) were known to have positive hormonal receptor. 10 patients received solely endocrine therapy and 13 patients received it concomitantly with chemotherapy at some point of their treatment. Of the 3 patients that were positive for Her- 2/neu over expression, 2 had concomitant ER+ expression, 1 received treatment with trastuzumab as adjuvant therapy, and the other one only received Tamoxifen. The third patient initially refused but upon relapsed, received trastuzumab, and then lapatinib was added at progression. The triple negative phenotype, i.e., ER−/PR−/HER2− was present in 10 patients who had reports of all markers.

Adverse events

All patients tolerated surgery well. From the 30 patients that receive chemotherapy either in neoadjuvant only (12), adjuvant only (14), or both (2), serious adverse events were documented in 14 patients (46 %). The most common complication was myelosuppression/neutropenic fever in 10 patients, including one death due to neutropenic fever. Two women developed Pneumocystis Carinii pneumonia during treatment. One patient had severe erosive candida esophagitis and another patient developed adult respiratory distress syndrome (ARDS) during treatment. These two patients developed acceleration of HIV disease to AIDS and death within 6 months of completing chemotherapy. Zoster infection was reported in 3 patients. One patient had elevated liver enzymes during treatment. Patients that were treated with endocrine therapy had no significant side effects. Overall, radiation therapy was well tolerated and there were no significant side effects documented.

Clinical outcomes

PFS

17 patients (35 %) relapsed and 8 more had progressive disease (17 %). The median PFS was 45 months, SE 36 (Fig. 1). The estimate PFS was 53 % at 3 years (95 % CI 36–67 %), 50 % at 5 years 95 % CI 34–64 %), and 40 % at 10 years (95 % CI 23–57 %). Of the 25 patients who had progressive disease or relapse, 5 (21 %) received chemotherapy, 8 (33 %) radiation, and 5 (21 %) received targeted therapy. Treatment for PD or Relapse by stage is also provided in Table 2. The PFS was associated with initial stage (Fig. 2), HR 3.59 (p < 0.000). 5 year PFS by Stage was 100 % for Stage I, 56 % for Stage II (95 % CI 31–75 %), 38 % for Stage III (95 % CI 16–60 %), and 0 % for Stage IV.

Fig. 1
figure 1

Probability of progression-free survival after breast cancer diagnosis in patients with HIV

Full size image
Fig. 2
figure 2

Probability of progression-free survival by initial stage after breast cancer diagnosis in patients with HIV

Full size image

OS

The median length of follow-up was 42 months (range from 1 to 243 months). Thirty-three patients (69 %) died during follow-up. The most common cause of death was breast cancer in 22 patients out of the 25 patients who had PD or relapsed. One of the patients is still following in clinic and the other two patients were lost to follow-up after disease progression. Six patients died from AIDS-related complications. One patient died from other medical condition and for four, the cause of death was unknown and death was confirmed by social security data. The median OS after breast cancer diagnosis was 52 months, SE 9 (Fig. 3). The estimate OS was 61 % at 3 years (95 % CI 46–73 %), 44 % at 5 years (95 % CI 29–58 %), and 28 % at 10 years (95 % CI 15–43 %).

Fig. 3
figure 3

Overall survival after breast cancer diagnosis in Patients with HIV

Full size image

Breast cancer-specific survival

The median breast cancer-specific survival was 117 months, SE 45. The estimate breast cancer-specific survival was 68 % at 3 years (95 % CI 52–80 %), 57 % at 5 years (95 % CI 40–70 %), and 47 % at 10 years (95 % CI 30–64 %). Breast cancer-specific OS was also associated with initial stage (Fig. 4), HR 3.00 (p < 0.001). 5 year Breast cancer-specific survival by Stage (Fig. 4) was 100 % for Stage I, 66 % for Stage II (95 % CI 39–83 %), 43 % for Stage III (95 % CI 17–67 %), and 0 % for Stage IV. The only patient with DCIS died from AIDS progression, and 5 year survival was unavailable.

Fig. 4
figure 4

Breast cancer-specific survival after breast cancer diagnosis in Patients with HIV

Full size image

Discussion

Comparing our 48 consecutive patients with HIV and breast cancer treated in a single institution to the historical data for the general breast cancer population, we found that our cohort was more likely to be young, and African American. The median age was 46 at the time of breast cancer diagnosis compared to 61 for the general population in the period 2006–2010. Our patients were younger than the median age for African Americans women without breast cancer (57 years) [3]. Only 33 % of our patients were older than 50 years compared to 79 % in the general population [27]. The occurrence of breast cancer in younger patients with HIV has been a common finding between reported case series [11–13, 19]; however, a recent study reported that these differences disappear after adjusting for differences in the populations at risk of HIV patients [28]. Half of our patients had ER-positive breast cancer and 6 % were HER-2 overexpressing.

Breast cancer was the main cause of death for the cohort and as anticipated, mortality of breast cancer was statistically related to the stage of the disease at presentation. In order to provide a context for our results, the overall survival of this cohort of HIV-infected women with breast cancer had higher breast cancer-specific mortality compared to the general American population and to the African American population. Breast cancer-specific OS at 5 years was 57 % in our HIV women compared to 89.2 % for general population reported by the SEER data base for the 2004–2010 period [3]. Caution should be used when interpreting these long-term survival rates because they represent patients who were diagnosed over the course of 15 years and therapy for both breast cancer and HIV has improved dramatically over this time period. [2]. In addition, these comparisons should be regarded with uncertainty as no direct assessment can be made between retrospective data with historical statistics as many things can be influencing the results, and no conclusion can be reached regarding behavior of breast cancer in HIV patients.

The competing risk and effect of HIV/AIDS in presentation, treatment complications and mortality are of importance interest in this population. Our cohort had similar distribution of CD4 count categories as the general population of women with HIV infection. There has been no association between HIV or CD4 count and stage at diagnosis for breast cancer patients [12, 15, 16]. In our series most women presented with stage II and III (83 %), which is higher than for the general population but more in line with what is seen in our safety net clinic population where the combined rate of Stage 2 and 3 is approximately 65 %.

The rationale for choosing specific therapeutic regimens in the HIV population has not been codified. The incidence of myelosuppression with chemotherapy appears to be higher in the HIV-infected population based on our initial report [18] where 5 out of 7 patients developed myelosuppression, and these findings have also been reported in others experience [20, 29]. In our updated and larger cohort, the incidence of serious adverse events was documented in 14 out of 30 patients that receive chemotherapy which reflects a proportionally decrease in the rate of complications. In addition to myelosuppression, zoster was a common complication as was thrush and vaginal candidiasis. In an attempt to minimize chemotherapy-related complication, we have adopted a number of strategies. We use weekly therapy rather than every 3 week therapy and we never use dose dense regimens since Burstein reported the occurrence of pneumocystisis carinii pneumonia in 2 women without HIV infection treated with dose dense AC-T [30]. We use ovarian suppression plus tamoxifen in women whose tumors are estrogen receptor positive instead of chemotherapy. We avoid steroid premedication when possible and we substitute Nab-Paclitaxel for Paclitaxel on occasion to avoid weekly steroid administration. We use growth factor support liberally and give prophylactic fluconazole, trimethoprim/sulfamethoxazole, and acyclovir to most women receiving chemotherapy. In patients with space occupying lesions of the brain we initiate therapy for toxoplasmosis. If there is no response to therapy for toxoplasmosis and the csf shows no Epstein Barr Virus (EBV) infection we consider that the patient has metastatic breast cancer to the brain and treat accordingly.

Drug interaction between chemotherapy and HAART is still an area of concern but limited data exist with breast cancer-specific regimens [26]. The earlier patients in our study did not receive HAART therapy and they had many more complications than the later patients in the study. We did not stop or modify HAART therapy for the patients receiving therapy. In the Soweto report [12], the experience of treating patients with very low CD4 count presented especial problems, as aggressive chemotherapy given without HAART could be rapidly fatal and holding chemotherapy until increased CD4 was not an option. They recommended initiating chemotherapy after initiating ART; however, no data were reported on complication, thus we cannot make any conclusions about the relationship between HAART and treatment on our patients

Our specific population may limit the external validity of our results. In 2010, 94,897 people were living with a diagnoses of HIV in Florida (a state with one of highest prevalence), of these 46, 248 (48 %) were African American despite making up only 16 % of the total population [31]. In Miami-Dade County, African American compromise 50 % of total HIV/AIDS population but represent 20 % of the general population [32]. These numbers are even starker in black women, where the black female to white female ratio of HIV/AIDS cases is 10 to 1 in Miami-Dade County.

The intrinsic worse prognosis of breast cancer in women of African descent where incidence is higher in younger women who have higher grade and hormone receptor-negative tumors compared to than those diagnosed in white women [33–35] is compounded by lack of access to health care. The fact that most of the women with breast cancer and HIV are of African descent may complicate the question of what role the HIV infection has on the innate lethality of the disease.

The main strength of this series is that it tracks the long-term experience of a single institution in the treatment of breast cancer in patients with HIV/AIDS. It provides useful although anecdotal experience on treatment strategies that have been developed over many years. With significant limitation in available cases even in a referral center as ours, observational studies constitute the best available data.

There are still many unanswered questions. Further studies to confirm the association between HIV status and survival with cancer, role, and limitations of HAART within selection of therapy, and optimal standard of care for this specific population persists as interesting areas of research. Both screening and adjuvant treatment have been associated with reduction of death from breast cancer in the general population, [1]; the application to this HIV subgroup remains to be optimized.

In conclusion, we have demonstrated in this updated series that treating breast cancer in HIV-infected woman is possible but remains a challenge to the oncologist. Surgery and endocrine-directed therapy remain well tolerated but standard-dose chemotherapy can have life-threatening side effects in HIV-infected patients. Hence, the benefits of adjuvant chemotherapy in this population have to be carefully weighed against the high incidence of complications and adequate choice of the least myelotoxic regimen. From our experience, if given with adequate growth factor support and prophylaxis, it can potentially improve outcomes. With increase life expectancy related to HIV with adequate HAART, the main cause of death in this women is expected be related to the breast cancer or treatment and adequate management will be key to survival but further research efforts are needed in this area.