Survival Impact and Cost‐Effectiveness of a Multidisciplinary Tumor Board for Breast Cancer in Mozambique, Sub‐Saharan Africa

Abstract

Background

Despite the international endorsement of multidisciplinary tumor boards (MTBs) for breast cancer care, implementation is suboptimal worldwide, and evidence regarding their effectiveness in developing countries is lacking. We assessed the impact on survival and the cost‐effectiveness of implementing an MTB in Mozambique, sub‐Saharan Africa.

Materials and Methods

This prospective cohort study included 205 patients with breast cancer diagnosed between January 2015 and August 2017 (98 before and 107 after MTB implementation), followed to November 2019. Pre‐ and post‐MTB implementation subcohorts were compared for clinical characteristics, treatments, and overall survival. We used hazard ratios and 95% confidence intervals (CI), computed by Cox proportional hazards regression. The impact of MTB implementation on the cost per quality‐adjusted life year (QALY) was estimated from the provider perspective.

Results

We found no significant differences between pre‐ and post‐MTB subcohorts regarding clinical characteristics or treatments received. Among patients with early breast cancer (stage 0–III; n = 163), the 3‐year overall survival was 48.0% (95% CI, 35.9–59.1) in the pre‐MTB and 73.0% (95% CI, 61.3–81.6) in the post‐MTB subcohort; adjusted hazard ratio, 0.47 (95% CI, 0.27–0.81). The absolute 3‐year mean cost increase was $119.83 per patient, and the incremental cost‐effectiveness ratio was $802.96 per QALY, corresponding to 1.6 times the gross domestic product of Mozambique.

Conclusion

The implementation of a MTB in Mozambique led to a 53% mortality decrease among patients with early breast cancer, and it was cost‐effective. These findings highlight the feasibility of implementing this strategy and the need for scaling‐up MTBs in developing countries, as a way to improve patient outcomes.

Implications for Practice

Currently, more than half of the deaths from breast cancer in the world occur in developing countries. Strategies that optimize care and that are adjusted for available resources are needed to improve the outcomes of patients with breast cancer in these regions. The discussion of cases at multidisciplinary tumor boards (MTBs) may improve survival outcomes, but implementation is suboptimal worldwide, and evidence regarding their effectiveness in developing countries is lacking. This study evaluated the impact of implementing an MTB on the care and survival of patients with breast cancer in Mozambique, sub‐Saharan Africa and its cost‐effectiveness in this low‐income setting.

Short abstract

This article evaluates the effect of implementing a multidisciplinary tumor board on the care and survival of patients with breast cancer in Mozambique, Sub‐Saharan Africa, including the cost‐effectiveness of this strategy in a low‐resource setting.

Introduction

Breast cancer is the most frequent cancer among women in the world, and its incidence rates have been increasing over the last decades, especially in developing countries [1]. Specifically in Mozambique, it is currently the second most incident cancer in Maputo City, with an age‐standardized rate of 15.5 per 100,000 women [2].

Breast cancer is still the most frequent cause of cancer death among women in developing regions, and age‐standardized mortality rates are higher in low‐ and medium‐income than in high‐income countries (14.9 vs. 11.6 per 100,000 women, respectively) [1]. In sub‐Saharan Africa, this is partly due to the large proportion of patients with stage III/IV disease at presentation, which was around 75% in a large pooled analysis [3]. Additionally, the percentage of women diagnosed with stage IV disease may also be high, reaching 70% in some series [3]. Moreover, the high prevalence of triple‐negative breast cancer (TNBC) [4] and the limited access to adequate diagnosis and care, namely pathology services, surgical care, radiotherapy, and systemic treatment, also contribute to patients’ dismal prognosis [5].

The discussion of cases at multidisciplinary tumor boards (MTBs) may be one of the strategies to improve outcomes in breast cancer. Most international oncology societies consider multidisciplinarity an essential part of breast cancer care, proposing that all patients should be presented in a MTB before initiating treatment [6]. Nonetheless, there is a wide gap between this universal recommendation and its effective implementation, even in developed countries. A French multicentric study showed that, although all breast cancer cases were presented in an MTB, only 64% were discussed before surgery, with significant differences between academic vs. nonacademic hospitals [7]. In a report using data from the U.S. SEER‐Medicare database, only 70% of patients had access to multidisciplinary care, and this proportion was even lower among those living in a rural area [8]. In an Australian study, only 60% of breast cancer cases diagnosed during 2010–2015 were discussed in MTBs [9]. In China, less than 2% of patients operated for breast cancer during 2006–2016 in a large cancer center were discussed in an MTB [10], and, in Taiwan, only 28% were presented in a multidisciplinary meeting [11]. In a recent survey conducted among members of the African Organisation for Research and Training in Cancer, 80% of the respondents reported they were involved in a multidisciplinary breast cancer team, but this is likely to be an overestimate, as there were only 37 respondents [12]. A total of 72% of the participants in a survey conducted in Arab countries reported having an MTB at their institution, although only 49% of these MTBs met on a weekly basis [13].

Data regarding the impact of breast cancer MTBs on survival is even scarcer. Although a large study from the U.K. and another from Germany demonstrated that multidisciplinary care improved survival among patients with breast cancer [14, 15], a population‐based study from Germany did not present such a survival benefit [16]. Studies from China and Taiwan showed a survival advantage with the implementation of multidisciplinary care for breast cancer [10, 11], but there are no data from other parts of the world. A study from the U.K. revealed that MTBs can be expensive resources [17], and they may be burdensome to some clinical departments, raising the question of their added value [18]. However, we could not find any study assessing the cost‐effectiveness of breast cancer MTBs.

Currently, more than half of the deaths from breast cancer in the world occur in developing countries. Therefore, it is paramount to develop strategies adjusted for the available resources that optimize care, translating into improved breast cancer survival. With this study, we aimed to assess the impact of implementing an MTB on the care and survival of patients with breast cancer in Mozambique. Additionally, its cost‐effectiveness was evaluated in this low‐resource setting.

Subjects, Materials, and Methods

Setting

Mozambique is a low‐income country from Eastern sub‐Saharan Africa with 30 million inhabitants. In 2018, the gross domestic product (GDP) was $499.00 per capita [19]. The average life expectancy is approximately 59 years [20], and 13.2% of adults aged 15–49 years are infected with HIV [21]. The country is now undergoing a surge of noncommunicable diseases, such as diabetes, cardiovascular disease, and cancer [22]. The density of physicians was 0.1/1000 inhabitants in 2017 [23], and the public health system, which is free‐of‐charge, is the largest health care provider. A total of 8.7% of the 2018 Mozambican state budget was allocated to the health sector, which is similar to other low‐income countries. It also relies on external resources (e.g., foreign donations, among others), which constituted 14% of the 2018 health budget [24]. The country has scarce resources for cancer care, namely in terms of facilities and health professionals. In 2015, there were only three pathology departments, two oncology units, and one surgery department with specialized surgical oncologists. Patients had free access to anthracyclines, cyclophosphamide, taxanes, methotrexate, and tamoxifen, although there were occasional interruptions in their supply. Trastuzumab and aromatase inhibitors were not provided by the public health system, and the first radiotherapy unit only opened in August 2019. Since 2013, the Calouste Gulbenkian Foundation has had a cancer care program for Mozambican health practitioners, in which they are trained in Portugal for short periods of time. Additionally, there are Cancer Registries in the Beira and Maputo Cities and in the Maputo Central Hospital [2, 25].

Study Design and Participants

We conducted the prospective Moza‐BC cohort study, which included consecutive patients with newly diagnosed breast cancer with a pathological diagnosis between January 2015 and August 2017 (supplemental online Fig. 1) [26]. Data on socio‐demographic characteristics, risk factors, HIV status, clinical and tumor characteristics, treatment, and survival of patients treated in the oncology unit of the Maputo Central Hospital were prospectively collected until July 2018. Survival data were updated through the Hospital's Cancer Registry, health records, and telephone interviews in November 2019.

Breast tissue samples were collected by surgical biopsy, breast surgery, and fine‐needle aspiration cytology and infrequently by core‐needle biopsies. Handling of breast specimens was standardized according to the American Society of Clinical Oncology/College of American Pathologists recommendations [27]. Expression of the estrogen receptor (ER), the progesterone receptor, and the overexpression and amplification of HER2 was assessed by immunohistochemistry on the primary tumor, as described in the literature [27, 28]. In cases in which the immunohistochemistry score of HER2 was 2+ (HER2‐equivocal), silver in situ hybridization was performed at the Centro Hospitalar São João, Portugal. Tumors were grouped according to the classic subtypes into ER‐positive/HER2‐negative, HER2‐positive (ER‐positive/HER2‐positive and ER‐negative/HER2‐positive), and TNBC (ER‐negative/ progesterone receptor‐negative/HER2‐negative).

Staging was defined by the AJCC TNM 7^th edition classification [29]. Clinical staging was performed using clinical examination, mammography or breast ultrasound, chest x‐ray, and abdominal ultrasound. Computed tomography scans were occasionally performed, but bone scans and breast magnetic resonance imaging were not available.

Multidisciplinary Tumor Board

The above‐mentioned training program in cancer care, along with the assembling and follow‐up of this cohort, raised awareness among Maputo Central Hospital physicians regarding the need for a better coordination of breast cancer care. Therefore, in March 2016, conditions were set to start an MTB for breast cancer, which was the first one to ever occur in Mozambique. The MTB meetings were performed on a weekly basis and lasted 1 hour, with the participation of at least one member from the surgery, oncology, pathology, and radiology departments and with trainees from these different specialties. Although radiotherapy was not yet available in Mozambique at the time, its implementation process was already ongoing, and there were Mozambican physicians being trained abroad in radiation oncology. Thus, a radiation oncologist who had recently completed his training was also present at the MTB and would indicate if the patient needed radiotherapy or not, even if only a small fraction of patients could have access to it abroad. A secretary recorded all decisions. With the creation of the MTB, patients’ navigation inside the Maputo Central Hospital changed from erratic to organized (Fig. 1). Every new breast cancer case is now presented in the MTB, and patients with relapse are often discussed as well. The list of cases is sent to the team a few days in advance and the medical oncologist usually does the presentation. Moreover, an effort has been done by the oncology unit team to ensure the adherence of patients to systemic therapy and follow‐up care, which led to an increase in the number of follow‐up visits from three to four visits per year during the first 3 years after diagnosis.

Figure 1.

Implementation of the MTB at the Maputo Central Hospital. Before March 2016 (left panel), upon suspicion of a breast tumor, the patient was referred to the surgical department of the Maputo Central Hospital (MCH), and the surgeon usually asked for a breast fine‐needle aspiration cytology biopsy to be performed at the pathology department. Alternatively, the patient had a biopsy performed outside the MCH or the surgeon performed a surgical biopsy at the MCH, but the tumor sample was analyzed and diagnosed at the pathology department. In those cases, the pathologist referred the patient to the surgical department. The surgeon then decided if the patient was to be operated on first or if she needed neoadjuvant chemotherapy. In the latter situation, the surgeon referred the patient to the oncology unit and the medical oncologist asked for staging exams to be performed at the radiology department. Upon completion of neoadjuvant chemotherapy, the patient was referred back to the surgical department. Following surgery, the patient was sent to the oncology unit again, to decide if she should also undergo adjuvant therapy and to receive long‐term follow‐up. After March 2016 (right panel), the referral scheme of patients from outside the MCH was maintained. However, following the pathological diagnosis of breast cancer, the patient is now directly referred to the oncology unit. She is often seen on the same day by the “on‐duty” oncologist, who asks for staging exams and schedules the MTB discussion within 1–2 weeks. Every new case of breast cancer is presented at the MTB regardless of stage, and patients with relapse are often discussed as well. The MTB team decides if the patient with early breast cancer is to receive neoadjuvant chemotherapy or surgery first. Following completion of neoadjuvant chemotherapy, the oncologist refers the patient to the surgical department, but also verbally informs the surgeon of this referral at the weekly MTB. After surgery, the patient's case is discussed again at the MTB to decide if she should also undergo adjuvant treatment and its type. Dashed arrows: infrequent interaction; full arrows: frequent interaction. Abbreviation: MTB, multidisciplinary tumor board.

Ethics

The National Health Bioethical Committee of Mozambique approved this study (226/CNBS/15). All participants provided written informed consent.

Statistical Analysis

Two subcohorts were defined according to the date of the MTB implementation (March 2016). Baseline patient, tumor, and treatment characteristics were compared between the two subcohorts using χ² test and Fisher's exact test for categorical variables and through adjusted odds ratios and 95% confidence intervals (CIs), computed using binomial logistic regression.

Overall survival was defined as time from diagnosis (date of pathological confirmation of breast cancer) until death by any cause. Among patients with early stage breast cancer (stage 0–III), disease‐free survival was defined as time from diagnosis until loco‐regional or distant relapse, second primary malignancy or death by any cause.

The median follow‐up duration was 41.0 months (maximum, 47 months) among patients on the pre‐MTB subcohort and 37.8 months (maximum, 43 months) among patients in the post‐MTB subcohort; thus, survival analyses were censored at 36 months and performed using the Kaplan‐Meier estimator. Comparisons of survival between the two subcohorts were accomplished through log‐rank tests and crude and adjusted hazard ratios (HRs) and corresponding 95% CIs, computed using Cox proportional hazards regression. Analyses were carried out in STATA version 15 (StataCorp, College Station, TX). All tests were two‐sided and a p value of <.05 was considered significant.

To estimate the impact of implementing a MTB on the cost per quality‐adjusted life year (QALY), measured over a 3‐year time horizon, a decision tree model was developed that integrated patient‐level data from this cohort study, including survival estimates (supplemental online Methods). All cost calculations assumed 2019 as the reference year. Costs determined in previous years were corrected for inflation using the GDP deflator, as proposed by the World Health Organization [30]; correction factors were extracted from the World Bank [31]. All costs determined in U.S. dollars (USD) or Euros were converted to Mozambique Meticais (MZN), using the exchange rates $1 = 63.8292 MZN and €1 = 70.5682 MZN, respectively; final results were converted back to USD, and are presented in both U.S. dollars and MZN. A detailed description of the cost‐effectiveness analysis methods can be found in the supplemental online Methods. The model was implemented using TreeAgePro 2018 R2.1 (Williamstown, MA).

Results

Patients’ median age was 48.0 years, and most were Black (98.0%), premenopausal (53.7%), and overweight or obese (54.1%) and presented with stage III/IV disease (74.1%). The prevalence of HIV infection was 25.4% (Table 1).

Table 1.

Patients’ characteristics and treatment according to date of diagnosis (pre‐ vs. post‐MTB implementation)

Characteristics	Pre‐MTB, n = 98	Post‐MTB, n = 107	p a	aOR b (95% CI)
Age in years, n (%)			.945
<40	26 (26.5)	28 (26.2)		1
40–49	29 (29.6)	28 (26.2)		0.81 (0.36–1.80)
50–59	21 (21.4)	25 (23.4)		1.10 (0.48–2.56)
≥60	22 (22.4)	26 (24.3)		1.07 (0.46–2.47)
Race, n (%)			.623
Black	97 (99.0	104 (97.2)		1
Other c	1 (1.0)	3 (2.8)		2.34 (0.22–25.41)
Education in years, n (%)			.618
0	18 (22.0)	23 (25.3)		1
1–4	14 (17.1)	11 (12.1)		0.63 (0.22–1.83)
>4	50 (61.0)	57 (62.6)		1.09 (0.45–2.67)
Missing	16	16
Place of residence, n (%)			.089
South (including Maputo)	80 (86.0)	96 (94.1)		1
Center/North	13 (14.0)	6 (5.9)		0.30 (0.10–0.88)
Missing	5	5
Menopausal status, n (%)			.889
Premenopausal	52 (53.1)	58 (54.2)		1
Postmenopausal	46 (46.9)	49 (45.8)		0.47 (0.13–1.70)
Body mass index, n (%)			.460
Under/normal weight, <25 kg/m2	33 (37.9)	44 (43.6)		1
Overweight/obese, ≥25 kg/m2	54 (62.1)	57 (56.4)		0.81 (0.42–1.59)
Missing	11	6
HIV status, n (%) d			.873
Negative/unknown	74 (75.5)	79 (73.8)		1
Positive	24 (24.5)	28 (26.2)		1.28 (0.65–2.53)
Classic subtype, n (%)			.028
ER+/HER2−	46 (46.9)	45 (42.1)		1
HER2+	22 (22.4)	22 (20.6)		0.98 (0.46–2.05)
TNBC	23 (23.5)	17 (15.9)		0.72 (0.33–1.56)
Unknown e	7 (7.1)	23 (21.5)		3.66 (1.39–9.64)
Stage at diagnosis, n (%)			.538
0‐II f	23 (23.5)	30 (28.0)		1
III	52 (53.1)	58 (54.2)		0.92 (0.45–1.86)
IV	23 (23.5)	19 (17.8)		0.53 (0.22–1.31)
Time from symptoms to diagnosis, n (%)			.598
<180 days	33 (54.1)	32 (50.0)		1
≥180 days	28 (45.9)	33 (50.8)		1.23 (0.57–2.64)
Missing	37	42
Time from diagnosis to treatment, n (%)			.324
<45 days	42 (44.7)	55 (51.9)		1
≥45 days	52 (55.3)	51 (48.1)		0.69 (0.38–1.25)
Missing	4	1
No treatment received: yes, n (%)	3 (3.1)	1 (0.9)	.351	0.32 (0.03–3.52)
Type of first treatment received, n (%)			.999
Surgery	24 (25.5)	26 (24.5)		1
Chemotherapy/endocrine therapy	70 (74.5)	80 (75.5)		0.94 (0.48–1.86)
Missing	1	0
Surgery (ever): yes, n (%) g	79 (80.6)	88 (82.2)	.858	1.44 (0.59–3.50)
Type of breast surgery, n (%) h			.257
Total mastectomy	75 (94.9)	79 (89.8)		1
Tumorectomy	4 (5.1)	9 (10.2)		2.44 (0.65–9.23)
Status of surgical margins, n (%)			.575
Clean	61 (88.4)	72 (92.3)		1
Positive	8 (11.6)	6 (7.7)		0.66 (0.20–2.19)
Missing i	10	10
Histological grade at surgery, n (%)			.422
1	16 (22.2)	20 (27.0)		1
2	35 (48.6)	28 (37.8)		0.62 (0.25–1.51)
3	21 (29.2)	26 (35.1)		1.22 (0.47–3.19)
Missing i	7	14
Axillary surgery: type, n (%) h			1.0
Sentinel lymph node biopsy	0	1 (1.3)
Axillary dissection	70 (100)	77 (98.7)
Axillary surgery: completeness, n (%) j			.423
Not done	9 (11.7)	9 (10.7)		0.93 (0.30–2.89)
Incomplete	23 (29.9)	18 (21.4)		0.58 (0.27–1.24)
Complete	45 (58.4)	57 (67.9)		1
Missing	2	4
Lymph node status, n (%)			.828
(y)pN0	18 (27.7)	22 (32.4)		1
(y)pN1	23 (35.4)	25 (36.8)		0.97 (0.39–2.43)
(y)pN2	13 (20.0)	13 (19.1)		0.94 (0.30–2.93)
(y)pN3	11 (16.9)	8 (11.8)		0.58 (0.17–2.04)
Missing i	14	20
Chemotherapy (ever): yes, n (%)	90 (91.8)	103 (96.3)	.237	2.36 (0.65–8.58)
Intent of first‐line CT, n (%) k			.174
Neoadjuvant only	8 (8.9)	20 (19.4)		1
Neoadjuvant + adjuvant	45 (50.0)	45 (43.7)		0.50 (0.19–1.31)
Adjuvant only	16 (17.8)	20 (19.4)		0.64 (0.21–1.96)
Palliative	21 (23.3)	18 (17.5)
Neoadjuvant CT: outcome, n (%)			.682
Same stage	19 (42.2)	22 (37.9)		1
Upstaging	8 (17.8)	8 (13.8)		0.72 (0.20–2.60)
Downstaging	18 (40.0)	28 (48.3)		1.35 (0.51–3.59)
Missing l	8	7
pCR rate after neoadjuvant CT, n (%)			.494
No pCR	48 (90.6)	56 (86.2)		1
pCR only in the breast (ypT0/is)	3 (5.7)	3 (4.6)		0.71 (0.11–4.75)
pCR in the breast and lymph nodes (ypT0/is, ypN0)	2 (3.8)	6 (9.2)		2.00 (0.34–11.96)
Type of first‐line CT regimen, n (%) k			.393
Anthracycline‐based only	44 (48.9)	45 (43.7)		1
Anthracyclines + taxanes based	40 (44.4)	55 (53.4)		1.22 (0.61–2.43)
Other m	6 (6.7)	3 (2.9)		0.37 (0.07–1.80)
First‐line CT dose‐intensity, n (%) k			.881
<85%	48 (57.8)	60 (59.4)		1
≥85%	35 (42.2)	41 (40.6)		0.95 (0.50–1.79)
Endocrine therapy (ever): yes, n (%)	50 (51.0)	54 (50.5)	.888	0.87 (0.45–1.66)
Radiotherapy (ever): yes, n (%)	2 (2.0)	7 (6.5)	.175	3.88 (0.74–20.27)

^a p value for the univariate comparison between the two subcohorts. Values in bold correspond to statistically significant differences.

^bAdjusted for age (<40 vs. 40–49 vs. 50–59 vs. ≥60 years), HIV status (negative/unknown vs. positive), stage at diagnosis (0–II vs. III vs. IV), and classic subtypes (ER+/HER2− vs. HER2+ vs. TNBC vs. unknown). The reference is the “pre‐MTB” implementation period. Values in bold correspond to statistically significant differences.

^cIncludes mixed and Asian race.

^dEight patients had unknown HIV status; among HIV‐positive patients, 38 (74%) had been previously diagnosed with HIV infection; the median time since HIV diagnosis was 3.93 years (range, 0.1–12.2); 48 (98.1%) patients were under antiretroviral treatment (ART) when starting chemotherapy, mostly with the TDF + 3TC + EFV regimen (33 patients); the median time under ART was 2.00 years (range, 0.1–12.2); the median CD4+ cells count was 439 cells/μL (range, 43–1,104), and 46 (90%) patients had a CD4+ cell count >200/μL.

^eBiomarker determination unknown due to carcinoma in situ or pathological complete response (i.e., no residual invasive tumor): n = 8 (n = 3 in the pre‐MTB and n = 5 in the post‐MTB subcohort); inadequate tissue fixation: n = 2 (0/2); unavailable samples: n = 20 (4/16).

^fAmong patients within the category 0–II stage, one patient had stage 0 and four patients had stage I.

^gIncludes 12 patients submitted to surgery outside the Maputo Central Hospital for whom complete pathological information regarding the surgical specimen is not available.

^hPatients submitted to a tumorectomy followed by a mastectomy were included in the "Mastectomy" group and those receiving a sentinel lymph node biopsy followed by an axillary dissection were included in the “Axillary dissection” group.

ⁱCases with missing data for surgical margins, histological grade, or lymph node status include patients submitted to surgery outside the Maputo Central Hospital for whom complete pathological information is not available regarding the surgical specimen; patients with pathological complete response; patients for whom axillary surgery was not performed; and patients for whom these assessments at the Maputo Central Hospital are missing.

^jAmong patients receiving any type of breast surgery (n = 166). Not done: not done or no isolated lymph nodes; incomplete: 1–5 isolated lymph nodes (in case of axillary lymph node dissection); complete: ≥6 isolated lymph nodes (in case of axillary lymph node dissection) or ≥ 1 isolated lymph nodes with ≤2 positive lymph nodes (in case of sentinel lymph node biopsy).

^kFirst line of chemotherapy that the patient received includes neoadjuvant, adjuvant, or palliative treatment. If the patient received part of chemotherapy as neoadjuvant (e.g., doxorubicin plus cyclophosphamide [AC] regimen) and another part as adjuvant chemotherapy (e.g., paclitaxel), the type of regimen and dose‐intensity refer to the entire scheme (neoadjuvant plus adjuvant).

^lCases with missing data regarding clinical staging or patient abandoned treatment.

^mIncludes taxane‐based chemotherapy (three patients), nonanthracycline/nontaxane‐based chemotherapy (five patients), and unknown regimen (one patient). The preferred anthracycline‐containing regimen was AC (cyclophosphamide 600 mg/m² and doxorubicin 60 mg/m² every 3 weeks), and the preferred taxane used was paclitaxel (175 mg/m² every three weeks); dose‐dense regimens were not used because of the unpredictable availability of granulocyte‐colony stimulating factors.

Abbreviations: aOR, adjusted odds ratio; CI, confidence interval; CT, chemotherapy; ER, estrogen receptor; MTB, multidisciplinary tumor board; pCR, pathological complete response; TNBC, triple‐negative breast cancer.

A total of 98 patients were included in the pre‐MTB subcohort and 107 patients were included in the post‐MTB subcohort. Patients in the post‐MTB subcohort had a higher chance of having an unknown breast cancer subtype and of living in the South of Mozambique, but there were no significant differences regarding other clinico‐pathological characteristics. Nearly half of the patients were diagnosed more than 6 months after the first symptoms, and half of them started treatment more than 45 days after diagnosis. The MTB implementation led to a slight increase in the use of chemotherapy (from 91.8% to 96.3%), the use of neoadjuvant chemotherapy only (from 8.9% to 19.4%), tumor downstaging with chemotherapy (from 40.0% to 48.3%), and the use of taxanes with anthracyclines (from 44.4% to 53.4%); however, these differences were not significant. There were also nonsignificant increases in the proportions of clean surgical margins (from 88.4% to 92.3%) and in complete axillary surgery (from 58.4% to 67.9%).

Overall Survival

A total of 23 women (11.2%) were lost to follow‐up, as their last contact with the hospital was more than 12 months before the survival cut‐off date, but they were included in the analysis. The 3‐year overall survival was 44.8% (95% CI, 34.2–54.8) in the pre‐MTB and 62.6% (95% CI, 51.8–71.6) in the post‐MTB subcohort (Fig. 2A). When restricting the analysis to patients with early breast cancer (stage 0–III), 3‐year overall survival was 48.0% (95% CI, 35.9–59.1) in the pre‐MTB and 73.0% (95% CI, 61.3–81.6) in the post‐MTB subcohort (Fig. 2B). Among patients with metastatic breast cancer, there was no survival benefit with the introduction of the MTB: median overall survival was 19.4 months (95% CI, 15.6–23.2) before and 13.6 months (95% CI, 9.2–17.9) after MTB implementation (Fig. 2C).

Figure 2.

Kaplan‐Meier curves for overall survival among all patients, patients with early breast cancer and patients with metastatic breast cancer; and for disease‐free survival among patients with early breast cancer. (A): Overall survival in the total population: 3‐year overall survival was 44.8% (95% confidence interval [CI], 34.2–54.8) in the pre‐MTB subcohort versus 62.6% (95% CI, 51.8–71.6) in the post‐MTB subcohort. (B): Overall survival among patients with early breast cancer (stage 0–III): 3‐year overall survival was 48.0% (95% CI, 35.9–59.1) in the pre‐MTB subcohort versus 73.0% (95% CI, 61.3–81.6) in the post‐MTB subcohort. (C): Overall survival among patients with metastatic breast cancer (stage IV): median overall survival was 19.4 months (95% CI, 15.6–23.2) in the pre‐MTB subcohort versus 13.6 months (95% CI, 9.2–17.9) in the post‐MTB subcohort. (D): Disease‐free survival among patients with early breast cancer (stage 0–III): 3‐year disease‐free survival was 41.7% (95% CI, 30.2–52.8) in the pre‐MTB subcohort versus 56.8% (95% CI, 45.3–66.8) in the post‐MTB subcohort. Abbreviation: MTB, multidisciplinary tumor board.

In the multivariable analysis, the adjusted HR for death in the post‐MTB vs. the pre‐MTB subcohort was 0.77 (95% CI, 0.49–1.19) in the overall population and 0.47 (95% CI, 0.27–0.81) among patients with early breast cancer (Fig. 3; supplemental online Tables 1–2). Factors significantly associated with overall survival in both the overall population and patients with early breast cancer were stage, tumor subtype, and type of first‐line chemotherapy regimen.

Figure 3.

Multivariable analyses of prognostic factors for overall survival (left panel) and disease‐free survival (right panel) among patients with early breast cancer (stage 0–III). All models included the following variables: MTB implementation period (pre vs. post), age (<40 vs. 40–49 vs. 50–59 vs. ≥60 years), HIV status (negative/unknown vs. positive), stage at diagnosis (0–II vs. III), and tumor subtype (ER+/HER2− vs. HER2+ vs. TNBC vs. unknown). ^¥Among patients receiving any type of breast surgery. Not done: not done or no isolated lymph nodes; incomplete: 1–5 isolated lymph nodes (in case of axillary lymph node dissection); complete: ≥6 isolated lymph nodes (in case of axillary lymph node dissection) or ≥1 isolated lymph nodes with ≤2 positive lymph nodes (in case of sentinel lymph node biopsy). ^§First‐line of chemotherapy that the patient received includes neoadjuvant and/or adjuvant treatment. If the patient received part of chemotherapy as neoadjuvant (e.g., doxorubicin plus cyclophosphamide [AC] regimen) and another part as adjuvant chemotherapy (e.g., paclitaxel), the type of regimen refers to the entire scheme (neoadjuvant plus adjuvant). Abbreviations: CI, confidence interval; CT, chemotherapy; ER, estrogen receptor; HR, hazard ratio; MTB, multidisciplinary tumor board; TNBC, triple‐negative breast cancer.

There was a benefit in MTB implementation among most subgroups of patients with early breast cancer (Fig. 4). There were apparently no benefits among those with lower education, HER2‐positive disease, grade 1 tumors, or receiving surgery as first treatment, but the low precision of the estimates precludes bold statements regarding stratum‐specific results or potential interactions.

Figure 4.

Multivariable analysis of overall survival in the post‐MTB versus pre‐MTB implementation periods within each subgroup of patients with early breast cancer (stage 0–III). All models included the following variables: MTB implementation period (pre vs. post), age (<40 vs. 40–49 vs. 50–59 vs. ≥60 years), HIV status (negative/unknown vs. positive), stage at diagnosis (0–II vs. III), and tumor subtype (ER+/HER2− vs. HER2+ vs. TNBC vs. unknown). ^¥Among patients receiving any type of breast surgery. Not done/incomplete: 0–5 isolated lymph nodes (in case of axillary lymph node dissection); complete: ≥6 isolated lymph nodes (in case of axillary lymph node dissection) or ≥ 1 isolated lymph nodes with ≤2 positive lymph nodes (in case of sentinel lymph node biopsy). ^§First‐line of chemotherapy that the patient received includes neoadjuvant and/or adjuvant treatment. If the patient received part of chemotherapy as neoadjuvant (e.g., doxorubicin plus cyclophosphamide [AC] regimen) and another part as adjuvant chemotherapy (e.g., paclitaxel), the type of regimen and dose‐intensity refer to the entire scheme (neoadjuvant plus adjuvant). Abbreviations: CI, confidence interval; CT, chemotherapy; ER, estrogen receptor; ET, endocrine therapy; HR, hazard ratio; MTB, multidisciplinary tumor board; TNBC, triple‐negative breast cancer.

Disease‐Free Survival

Among patients with early breast cancer, 48% had a disease‐free survival event: 18% had loco‐regional relapse, 8% had distant relapse, 6% had both loco‐regional and distant relapse, 1% had a second primary cancer, and 16% died. The 3‐year disease‐free survival was 41.7% (95% CI, 30.2–52.8) in the pre‐MTB and 56.8% (95% CI, 45.3–66.8) in the post‐MTB subcohort (Fig. 2D). The proportion of patients with loco‐regional relapses (with or without concomitant distant relapse) or death as the first disease‐free survival event was numerically higher in the pre‐MTB vs. the post‐MTB subcohort (29% vs. 18%; and 23% vs. 10%, respectively), but these differences were not statistically significant (p = .07). In the multivariable analysis, the adjusted HR for relapse or death was 0.72 (95% CI, 0.46–1.13) in the post‐MTB compared with the pre‐MTB subcohort (Fig. 3; supplemental online Table 3). Tumor characteristics (subtype, lymph node status, and stage), quality of surgery (surgical margins status and axillary surgery completeness), and type of first‐line chemotherapy regimen were significantly associated with disease‐free survival.

Cost‐Effectiveness Analysis

The absolute 3‐year mean cost increase of implementing a MTB was $119.83 per patient (7648.42 MZN per patient). The incremental cost‐effectiveness ratio was $802.96 per QALY (51252.57 MZN per QALY), corresponding to 1.6 times the GDP of Mozambique (Table 2). Assuming a willingness‐to‐pay threshold of three times the national GDP (1497.00 USD), MTB implementation is a cost‐effective measure in this setting. Sensitivity analyses showed similar findings (supplemental online Table 4).

Table 2.

Costs, benefits, costs per unit of benefit and incremental cost‐effectiveness ratios, by strategy (no multidisciplinary tumor board [MTB] vs. MTB), in a provider perspective, over a 3‐year time horizon

Strategy	Probability of being alive without relapse at 3 years	Probability of being alive with relapse at 3 years	Mean 3‐year cost per patient treated a	Mean 3‐year incremental cost per patient treated a	Mean QALYs per patient	ICER (cost/QALY) b
No MTB (usual care)	0.4166	0.0635	$993.1803; 63,393.9057 MZN		1.46225
MTB	0.5681	0.1615	$1,113.0067; 71,042.3278 MZN	$119.8264; 7,648.4221 MZN	1.61148	$802.9646; 51,252.5772 MZN

^aFuture health benefits and costs were discounted at 3% annually, assuming a time‐horizon of 3‐years, based on the World Health Organization recommendations [30].

^bThe conversion between U.S. dollars (USD) and MZN was performed using the exchange rate 1 USD = 63.8292 MZN.

Abbreviations: ICER, incremental cost‐effectiveness ratio; MTB, multidisciplinary tumor board; MZN, Mozambique Meticais; QALY, quality‐adjusted life year.

Discussion

To the best of our knowledge, this is the first study reporting the survival benefit and cost‐effectiveness analysis of a MTB for breast cancer in a low‐income country and in Africa. Our prospective cohort study showed that the implementation of a MTB, after the training of a few key health professionals, led to a significant reduction in mortality among patients with early breast cancer. This benefit remained significant after adjusting for other prognostic factors. Given the baseline poor prognosis of these patients, the absolute improvement in 3‐year overall survival among patients with early breast cancer was 25.0% (from 48.0% to 73.0%). Moreover, we have demonstrated that this is a cost‐effective intervention.

Interestingly, time from diagnosis until treatment or the proportion of each treatment received were similar in both the pre‐MTB and post‐MTB subcohorts. Although not statistically significant, there were improvements in the quality of care (i.e., clean surgical margins, use of neoadjuvant chemotherapy, among others). Thus, the survival benefit brought by the MTB might be from the added effect of each of these small improvements and by the truly integrated multidisciplinary care of patients, generating high‐value clinical practices and better patient navigation within the system. This is suggested by the fact that patients receiving neoadjuvant chemotherapy benefited more from MTB implementation than those receiving surgery first, probably because those patients would have had a more difficult navigation in the pre‐MTB era, potentially leading to delays between treatments and a worse outcome.

In contrast, the absence of survival benefit among patients with stage IV breast cancer may be related to their baseline poor prognosis and the low availability of effective systemic therapies. This highlights the need to diagnose breast tumors at an earlier stage and to provide more effective treatment strategies to women with de novo metastatic disease in low‐resource settings.

Studies assessing the composition, functioning, and impact of MTBs in different cancer types in high‐income countries have heterogeneous results in terms of survival impact and cost‐effectiveness [14, 15, 16, 32, 33]. These discrepancies may be due to different study designs, patient population, or MTB models. For instance, differences between two large studies in the U.K. [14] and Germany [16] in terms of inclusion periods (1990–2000 vs. 2004–2010, respectively), median follow‐up time (7.0 vs. 3.3 years), and diverse multidisciplinary care models may explain why the former showed a survival advantage of this intervention and the latter did not. In developing countries, data regarding MTBs are much scarcer, especially in sub‐Saharan Africa. A study from Botswana showed a reduction in the time from biopsy to start of radiation therapy among patients with cervical cancer [34]. After the implementation of gastro‐intestinal MTBs in Saudi Arabia and Ethiopia, adherence to evidence‐based guidelines improved [35, 36]. Unfortunately, these studies did not provide survival data.

One of the major strengths of the present analysis is being based in the prospective follow‐up of a cohort, unlike most breast cancer studies in sub‐Saharan African, which are usually retrospective [5, 37, 38, 39]. We collected information regarding the entire spectrum of cancer care, from the presentation of symptoms to diagnosis, comorbidities, treatment, and survival outcomes. There were also fewer losses to follow‐up (11.2%) than in previous larger retrospective studies (losses of 22%–53%) [38, 39, 40]. This reflects the efforts by the oncology unit staff to improve follow‐up completeness, by recording multiple telephone numbers per patient and actively contacting those who missed appointments.

During the study, there was no improvement in the access to certain drugs (e.g., trastuzumab remained unavailable) or to other treatments (e.g., radiotherapy), thus not confounding the effect estimated for the introduction of the MTB. There were also no significant differences between patients regarding prognostic factors, such as age, stage, HIV status, or proportion of tumor subtypes.

However, our study has limitations that should be acknowledged. We compared patients diagnosed during the 15 months preceding the MTB implementation to patients diagnosed in the 18 months that followed it. Given the time proximity between the two subcohorts, there are patients in the pre‐MTB subcohort who may have benefited from changes in cancer care brought by the training of professionals and MTB implementation and who may have been discussed in the MTB upon relapse. However, this might have contributed to underestimate the effectiveness of MTB implementation, and thus, this study provides conservative estimates of its effect.

There were no differences in disease‐free survival between the two subcohorts, but this might be related to the timing of relapse diagnosis. Patients in the pre‐MTB were not followed as closely as those in the post‐MTB subcohort, and, therefore, documentation of relapse may have occurred later compared with patients in the post‐MTB subcohort. This is reinforced by the fact that most disease‐free survival events consisted of loco‐regional disease, a type of relapse that is easily recognized by both patients and doctors, when compared with distant metastases. Thus, a more frequent clinical examination can detect loco‐regional relapse earlier, without the need for imaging exams. In addition, the proportion of patients with death as the first disease‐free survival event was numerically higher in the pre‐MTB vs. the post‐MTB subcohort, suggesting that these patients probably had an earlier relapse that was not detected by the physician, possibly because of the less close follow‐up in the pre‐MTB subcohort. Even if there were no significant differences in disease‐free survival, there was a significant benefit in overall survival, which is considered to be a more objective survival measure.

Surprisingly, there was a higher proportion of patients with tumors with unknown subtype in the post‐MTB compared with the pre‐MTB subcohort (21.5% vs. 7.1%, respectively). The most common reason was the unavailability of samples (i.e., there were no cell blocks from fine‐needle aspiration biopsies, core‐needle biopsies, or surgical specimens available to perform subtype determination). This could be because some patients were diagnosed and/or received surgery outside the Maputo Central Hospital and that there were fewer patients with stage IV disease operated to their breast tumor and a slightly higher number of patients with pathological complete response (i.e., without residual invasive tumor) in the post‐MTB subcohort.

In this study, we only analyzed patients treated and followed at a single center. Nonetheless, our population had many of the characteristics demonstrated in other sub‐Saharan African studies, such as a high proportion of stage III/IV at presentation [3], a delay of several months between first symptoms and diagnosis [41], a high proportion of TNBC [4], a high prevalence of HIV infection [40], and low survival [42]. This supports the possibility of generalizing our findings to other hospitals and settings in this region.

We showed that improving the continuum of breast cancer care leads to significant gains in survival among patients with early breast cancer, who are potentially curable. These results reflect (a) the training of key professionals in specialized breast cancer care; (b) a better coordination of care, through the implementation of a MTB; and (c) an active follow‐up of patients, leading to an improved adherence to treatment. Some of these solutions do not need significant extra funding to be implemented, but just a better organizational model.

We demonstrated that beyond traditional clinico‐pathological factors, not only does the performance of surgery or chemotherapy matter, but its quality is also paramount to improve survival. This is another argument to adequately train surgeons and oncologists in the best cancer care practices, as it can improve survival without a large increase in cost. Additionally, strategies to increase access to taxanes and chemotherapy dose‐intensity, such as an adequate supply of drugs and supportive treatment, should also be considered. These could lead to an increase in the proportion of patients with pathological complete response after neoadjuvant chemotherapy, which we have previously shown to be low across all tumor subtypes [26].

The success of this first MTB in Mozambique has recently led to the formal endorsement of multidisciplinarity as an essential part of cancer care, both in the Mozambican Breast Cancer Guidelines and in the Mozambican Cancer Control Plan 2019–2029 [43, 44]. Moreover, there are now separate MTBs for thoracic conditions, gynecological cancer, head and neck cancer, and esophageal cancer at the Maputo Central Hospital.

We will continue to update the survival follow‐up of this cohort using data from the Hospital's Cancer Registry. Yet, given the early and wide separation of the overall survival curves among patients with early breast cancer, it is expected that the survival benefit brought by the MTB will be maintained in the future.

We acknowledge the wide variations in available resources for cancer care among sub‐Saharan African countries and regions. Nonetheless, our proposed MTB model is simple, easy to implement, and cost‐effective, so we believe that it could be scaled up to other countries and settings, in which MTB implementation remains suboptimal.

Future research should also focus on ways to improve the effectiveness of MTBs in low‐resource settings, by identifying the processes most associated with survival gain. Moreover, the selection of patients that benefit the most from multidisciplinarity is also warranted. This future knowledge could also be used to improve the effectiveness of MTBs in high‐income countries.

Conclusion

Within this prospective cohort study in Mozambique, we showed that the implementation of an MTB led to a 53% mortality decrease among patients with early breast cancer. This intervention led to an absolute gain in 3‐year overall survival of 25.0% and was cost‐effective. These findings highlight the feasibility of implementing this strategy and the need for expanding the use of MTBs in developing countries as a way to improve patient outcomes. Given the high proportion of women with stage IV disease in our cohort, we also point to the need of diagnosing breast cancer at an earlier stage, to optimize patient outcomes.

Author Contributions

Conception/design: Mariana Brandão, Assucena Guisseve, Carla Carrilho, Nuno Lunet

Provision of study material or patients: Mariana Brandão, Assucena Guisseve, Genoveva Bata, Josefo Ferro, Carlos Garcia, Clésio Zaqueu, Astrilde Jamisse, Jotamo Come, Otília Soares, Alberto Gudo‐Morais, Satish Tulsidás, Carla Carrilho

Collection and/or assembly of data: Mariana Brandão, Assucena Guisseve, Genoveva Bata, Josefo Ferro, Carlos Garcia, Clésio Zaqueu, Cesaltina Lorenzoni, Dina Leitão, Otília Soares, Alberto Gudo‐Morais, Fernando Schmitt, Satish Tulsidás, Carla Carrilho

Data analysis and interpretation: Mariana Brandão, Assucena Guisseve, João Firmino‐Machado, Carla Carrilho, Nuno Lunet

Manuscript writing: Mariana Brandão, Assucena Guisseve

Final approval of manuscript: Mariana Brandão, Assucena Guisseve, Genoveva Bata, João Firmino‐Machado, Matos Alberto Josefo Ferro, Carlos Garcia, Clésio Zaqueu, Astrilde Jamisse, Cesaltina Lorenzoni, Martine Piccart‐Gebhart Dina Leitão, Jotamo Come, Otília Soares, Alberto Gudo‐Morais, Fernando Schmitt, Satish Tulsidás, Carla Carrilho, Nuno Lunet

Disclosures

Mariana Brandão: Radius, AstraZeneca, Eli Lilly & Co, Merck Sharpe & Dohme, GlaxoSmithKline/Novartis, Roche/GNE, Synthon, Servier, Pfizer (RF [Institution]). Roche/GNE (Other‐travel grant, H); Martine Piccart‐Gebhart: AstraZeneca, Camel‐IDS, Crescendo Biologics, Debiopharm, G1 Therapeutics, Genentech, Huya, Immunomedics, Eli Lilly & Co, Menarini, Merck Sharpe & Dohme, Novartis, Odonate, Periphagen, Pfizer, Roche, Seattle Genetics (C/A, H), Oncolytics (SAB), IJB from Radius, AstraZeneca, Lilly, Merck Sharp & Dohme, GlaxoSmithKline/Novartis, Roche/GNE, Synthon, Servier, Pfizer (RF). The other authors indicated no financial relationships.

(C/A) Consulting/advisory relationship; (RF) Research funding; (E) Employment; (ET) Expert testimony; (H) Honoraria received; (OI) Ownership interests; (IP) Intellectual property rights/inventor/patent holder; (SAB) Scientific advisory board

Supporting information

See http://www.TheOncologist.com for supplemental material available online.

Supplemental Figure 1

Appendix S1 Supporting Information.

Supplementary Table S1 Multivariable analysis of prognostic factors for overall survival among the total population

Supplementary Table S2 Multivariable analysis of prognostic factors for overall survival among patients with early stage breast cancer (stage 0‐III)

Supplementary Table S3 Multivariable analysis of prognostic factors for disease‐free survival among patients with early stage breast cancer (stage 0‐III)

Supplementary Table S4 Cost‐effectiveness sensitivity analyses^a