A study on patients’ selection for BRCA1 and BRCA2 mutations testing by different models in Libyan women with breast cancer

ABSTRACT

Introduction

The BRCA mutation spectrum of familial breast cancer in Libya remains unknown. Several genetic models developed to predict the probability of BRCA1/2 mutations have not been applied in Libya, where the NCCN criteria are used for highly penetrating breast cancer susceptibility genes. We aimed to predict BRCA1/2 mutation probability in familial breast cancer and genetic testing eligibility using BOADICEA and BRCAPRO models and NCCN criteria.

Methods

BRCA1/2 mutations were retrospectively predicted in 62 women with familial breast cancer between 2018 and 2021. Logistic regression, ROC analysis, and area under the curve were used to compare NCCN referral criteria with the BRCAPRO and BOADICEA scores.

Results

Thirty-two out of 62 breast cancer patients (51.6%), with a mean age of 43.5 ± 8 years, were predicted as BRCA mutation carriers by both models. BRCAPRO predicted BRCA1 and BRCA2 mutations in 27.4% and 41.9% of the women, respectively. BOADICEA predicted 8% for BRCA1 and 29% for BRCA2. At least one NCCN criterion was met by 50/62 women (80.6%). Three criteria were statistically significant predictors in BRCAPRO and BOADICEA: breast cancer at age ≤50 years with one or more close blood relatives with breast cancer, breast cancer patient with a close relative of male breast cancer, and triple-negative breast cancer. For the three respective criteria, sensitivity was 0.78, 0.89, and 0.75, specificity was 0.33, 0.39, and 0.22, area under curve was 0.72, 0.75, and 0.76, positive predictive value was 55%, 61%, and 51%, and negative predictive value was 58.5%, 77%, and 45%.

Conclusions

Our study highlights that certain aspects of the NCCN criteria demonstrate variations in significance when compared to the BRCAPR and BOADICEA models. For the first time, these models were used to predict BRCA mutations in Libyan women, and our finding indicates that these models are promising for improving genetic testing decision-making.

KEYWORDS:

• Breast cancer
• NCCN criteria
• BRCAPRO
• BOADICEA
• BRCA1
• BRCA2

1. Introduction

Breast cancer (BC) is the most common cancer worldwide [1]. Although the incidence rates in developed countries are 88% higher than in developing countries, the incidence rates in developing countries have increased rapidly [1]. North African countries are experiencing some of the most rapid increases, including Mauritania, Morocco, Algeria, Tunisia, and Libya [1,2]. In Libya, the incidence is 18.8 new cases per 100,000 women annually. Most of the patients have advanced disease and are often younger than in Europe, as reported elsewhere in North Africa [3]. Although the rapid rise in BC incidence rates in North African countries could be attributed to lifestyle and environmental factors, the onset at age of ≤40 year and the higher grade suggest the influence of genetic factors, such as mutations in breast cancer susceptibility genes (BRCA1 and BRCA2) [4]. While hereditary genetic factors contribute to 5–10% of all breast cancers globally [5], their specific roles in North African populations remain largely unexplored [6,7].

BRCA1/2 gene mutations account for up to 25% of familial cases [5]. The lifetime cancer risk in BRCA1/2 mutation carriers is roughly 80–85% for female breast cancer, 12–40% for bilateral breast cancer for BRCA1, and 6–10% for male breast cancer [8–13]. These statistics are based on research on European and American populations, and they likely differ in other populations.

Clinical standards [14,15] and predictive models of BRCA1/2 mutation carriers [16–19] have been developed, and data have accumulated. This has resulted in the development of various models that can more precisely estimate probabilities based on both genetic and empirical models that calculate the probability of mutation by using predictor variables derived from cancer family history. The BRCA carrier prediction model (BRCAPRO) software [18,20] and the Breast and Ovarian Analysis of Disease Incidence and Carrier Estimation Algorithm (BOADICEA) are examples of such tools [21,22]. Both BOADICEA and BRCAPRO models have been established for BRCA1/2 mutation prediction. While primarily developed and validated in European populations, preliminary studies suggest their potential applicability in diverse genetic backgrounds. Furthermore, BRCAPRO is based on mutation rates and penetrance in women of Ashkenazi Jewish and European ancestry [8,23].

The high cost and difficult interpretation of BRCA1/2 genetic testing, particularly for families with a low preexisting risk of having a mutation, underscore the importance of estimating the likelihood of a family carrying a BRCA1/2 gene mutation before a genetic test is done [24]. This becomes even more crucial in regions like Libya, which lacks readily available genetic testing for cancer patients, including those with breast cancer [25]. Libyan patients have to seek laboratories abroad. In view of the limitations in diagnostic facilities in Libya, risk assessment and genetic test recommendations for familial breast and ovarian cancer patients rely solely on the National Comprehensive Cancer Network (NCCN) criteria for highly penetrating breast cancer susceptibility genes. The NCCN recommends BRCA genetic testing for patients with BC who meet one of the following criteria: BC women diagnosed at age ≤45 years, BC at any age with ≥1 close male blood relative with BC, triple negative BC women at any age, BC women diagnosed at age ≥51 years with ≥1 close blood relative with BC at age ≤50 years, and BC women diagnosed between the ages of 46 and 50 years with any of the following: 1) multiple primary BC; 2) ≥1 close blood relative with breast, ovarian, pancreatic, or prostate cancer at any age [26].

Though the NCCN criteria provide valuable guidance, they lack specificity for accurately excluding people with low-priority risks [27]. To the best of our knowledge, BRCA1/2 mutation prediction models have not been applied to people of North African ethnicity. Therefore, we aimed to retrospectively predict BRCA1/2 mutation probability in Libyan women with BC by using both BOADICEA and BRCAPRO models and to compare how the NCCN high-risk assessment criteria perform in referring BC patients for genetic evaluation in comparison to the two risk assessment models.

2. Methods

2.1. Patients and criteria

The study included 62 unrelated women with BC who were followed up at the outpatient clinic of the National Cancer Institute in Sabratha, Libya from 2018 to 2021. The small sample size is a result of strict selection criteria. The patients were selected based on a family history of breast, ovarian, or other cancer in one or more first-, second-, or third-degree relatives, regardless of the age of onset. Data were collected from the patients in a self-administered questionnaire during their visits to the clinic. The medical records of the patients were reviewed.

2.2. Study tools

To calculate the probability of a patient being a BRCA1/2 mutation carrier, we utilized two freely accessible software programs: BRCAPRO CancerGene software program (v6, Bayes Mendel R package) (http://www4.utsouthwestern.edu/breasthealth/cagene) and BOADICEA CanRisk v6 (https://www.canrisk.org/). The BRCAPRO model uses statistical ideas that go back to Bayes and Mendel [20]. It analyzes data from patients and all their relatives, including their ages at diagnosis, current age, age at death, ethnicity, BC markers, such as the ER (estrogen receptor) scores, PR (progesterone receptor), and HER2 (human epidermal growth factor receptor-2), as well as other risk factors such as the woman’s age at first live birth, and mammographic density. The calculations make use of information on the occurrence of breast, ovarian, and other cancers among first- and second-degree relatives.

The BOADICEA model uses information such as lifestyle, women's health, number and sex of children, breast screening, mammographic density, hormone receptors, including ER, PR, and HER2, reproductive factors, and medical and family history. The medical history of the patients included age at diagnosis of breast or ovarian cancer. The family history included age at diagnosis of breast, ovarian, pancreatic, or prostate cancer in first- or second-degree relatives. In our study, the ethnicity of all the patients was marked as “unknown” in BRCAPRO and as “other” in BOADICEA because of insufficient research on the ethnic groups living in North Africa. Based on the probability calculation scores in the two models, ≥10% was considered a high-risk subgroup that would benefit from genetic testing [28].

As reported in St. Gallen International Breast Cancer Conference 2011 [29], patients’ tumors were classified into four molecular subtypes according to the following definitions: luminal A (ER+ and/or PR+, Ki67 low and HER2-), luminal B (ER+ and/or PR+, Ki67 high and/or HER2+), HER2-positive (ER-, PR-, and HER2+) and triple-negative (ER-, PR-, HER2-). Medical records were used to obtain ER, PR, HER2, and Ki67 immunohistochemical expression levels. ER and PR were recorded in categories (positive vs. negative). Patients with ˃ 20% positive nuclei were identified as having high Ki67 expression, while those with ≤20% positive nuclei were identified as having low Ki67 expression. HER2 was considered positive if it scored 3+ on immunohistochemistry [30]. All other scores (0 to 2+) were considered negative unless fluorescence in situ hybridization (FISH) was done for equivocal scores (2+).

Following the NCCN guidelines (v2.2022) [26], patients were assessed for referral to genetic risk evaluation based on the following criteria:

• BC women diagnosed aged ≤45 years;

• BC women diagnosed at age ≤50 years with ≥1 close blood relative with breast or ovarian cancer at age ≤50;

• BC women between at age 46 and 50 years together with primary breast or ovarian cancer, or with ≥1 close blood relative with pancreatic or prostate cancer at any age;

• BC women diagnosed at any age with ≥1 close blood relative diagnosed with male breast cancer at any age;

• Triple-negative BC women diagnosed at age ≤60 years.

NCCN does not provide a percent probability of mutation, whereas the models calculate carrier probability scores for the BRCA1 and BRCA2 genes separately.

We compared each of the NCCN referral criteria to the BRCAPRO and BOADICEA scores of our breast cancer patients to determine how each of the NCCN criteria performed.

2.3. Statistical analysis

We summarized the clinical characteristics of the study participants as medians for continuous variables and frequencies for categorical variables. Logistic regression was used in conjunction with receiver operating characteristic (ROC) analysis to determine which NCCN criteria were statistically significant predictors of high-risk patients, and the BRCAPRO and/or BOADICEA were used as the reference standards to evaluate the sensitivity and specificity of the patients’ criteria. To evaluate the sensitivity and specificity of individual NCCN criteria, we considered patients with high-risk predictions from both BRCAPRO and/or BOADICEA (reference-positive group) and patients who met the NCCN tested criterion (test-positive group) as true positives. The area under the curve (AUC), a general indicator of accuracy, was calculated using ROC curves, and the calculations were made for both positive predictive values (PPV) and negative predictive values (NPV). A perfect test has an area of 1, whereas an area of 0.5 means that the test is useless. We examined each of the NCCN referral criteria with reference to the BRCAPRO and BOADICEA results of our breast cancer patients to see how each of the NCCN criteria performed. The clinical significance of each NCCN criterion was determined by BRCAPRO and/or BOADICEA scoring. All statistical analyzes were performed with IBM SPSS Statistics v20. The study power was calculated using G*Power statistics, and the results show an actual power of 0.86 with a 31-total sample size.

3. Results

The clinical characteristics of the 62 BC women who participated in the study are summarized in Tables 1 to 4. Forty-eight (77%) of them were ≤50 years old at diagnosis, with an average of 44.8 years (range: 20–67). Thirty-eight women (61.2%) had a family history of BC in first-degree relatives and 13 (21%) in second-degree relatives. One woman was diagnosed with double primary breast cancer, and another with both breast and ovarian cancer. The histological subtypes were invasive ductal carcinoma in 59/62 women (95.2%), and the other three had medullary, lobular, and ductal carcinoma in situ, respectively.

Table 1. Clinical and pathological characteristics of patients.

Characteristic	n (%)
Age ≤ 50 years	48 (77.4)
Unilateral breast cancer	60 (96.8)
Additionally, primary breast or ovarian cancer	2 (3.2)
Histological type of breast cancer Invasive ductal carcinoma Medullary carcinoma Invasive lobular carcinoma Ductal carcinoma in situ	59 (95.2) 1 (1.6) 1 (1.6) 1 (1.6)
Family history of breast cancer in: 1^st-degree relatives 2^nd-degree relatives 3^rd-degree relatives Family history of ovarian cancer in: 1^st-degree relatives Family history of prostate or pancreatic cancer Family history of endometrial cancer in: 1^st-degree relatives	38 (61.2) 13 (21) 8 (13) 1 (1.6) 4 (6.4) 4 (6.4)

Table 2. Distribution of molecular subtype of breast cancer by age groups and menopausal status.

		Molecular subtype
		Premenopause				Postmenopause
Age	N	Luminal A N (%)	Luminal B N (%)	HER2- Positive N (%)	Triple- Negative N (%)	Luminal A N (%)	Luminal B N (%)	HER2- Positive N (%)	Triple- Negative N (%)
≥ 50 years	48	3 (4.8)	24 (38.7)	3 (4.8)	2 (3.2)	4 (6.5)	8 (12.9)	0	4 (6.5)
≤ 51 years	14	(1.6)1	1 (1.6)	1 (1.6)	0	3 (4.8)	6 (9.7)	0	2 (3.2)
Total	62	4(6.5)	25 (40.3)	4 (6.5)	2 (3.2)	7 (11.3)	14 (22.6)	0	6 (9.7)

Number of patients N.

Table 3. BRCAPRO and BODICEA models against NCCN criteria for BRCA1/2 gene mutation risk prediction.

NCCN Criteria A woman with BC diagnosis meeting any of the following:	No. of patients	BRCAPRO patients high score (≥ 10)			BODICEA patients high score (≥ 10)						Number of patients with high score (≥ 10) by BRCAPRO or BOADICEA		P value for the high score (≥ 10) by BRCAPRO or BOADICEA
		Patient ID	BRCA1 Score (N=17)	Patient ID	BRCA2 score (N = 26)	Patient ID	BRCA1 Score N= 5))	Patient ID		BRCA2 Score (N=18)	BRCA1 Score (N =18)	BRCA2 score (N = 29)	BRCA1	BRCA2
BC at age ≤ 50 and ≥ 1 close blood relative with breast or ovarian cancer at age ≤ 50	35	14 15 18 19 20 21 24 25 27 28 39 44 48 62	28.9 21.2 17.9 23.3 12.1 14.8 50.0 36.1 21.7 12.3 20.4 38.8 42.6 18.9	3 14 15 17 18 19 20 21 23 25 27 28 30 32 33 35 44 47 48 49 57 62	10.1 30.6 32.7 80.9 48.8 55.0 54.4 10.2 34.2 42.2 44.7 11.6 25.7 67.2 14.3 45.0 53.0 62.0 25.6 30.6 90.8 26.9	9 18 19 25	18.3 25.7 25.7 75.7	14 15 17 18 19 20 23 30 32 37 44 47 57 62		29.1 19.2 61.5 32.1 32.1 63.0 18.9 29.1 23.0 29.1 29.6 50.2 63.4 17.8	15	23	.011 *	.005 *
At age 45≤	10	58	12.70	58	13.200						1	1	.178	.031
Double primary BC/OC	2	25	36.1	25 45	42.1 15.0	25	75.7		45	10.1	1	2	.487	.999
Male BC relative	9	20 27 44 62	12.1 21.7 38.8 19.0	17 20 23 27 44 47 49 57 62	80.9 54.4 34.2 44.7 53.0 62.0 30.6 90.8 26.9				17 20 23 44 47 57	61.5 32.1 18.9 29.6 50.2 63.4	4	9	.279	.023 *
≥ 1 relative with prostate or pancreatic cancer	4			32	67.2				32	23.00		1	.999	.385
Triple-negative (ER, PR, Her2-) BC diagnosed ≤60 years old	8	13 39 44 58 62	12.3 20.4 38.8 12.7 18.9	44 58 62	53.0 13.2 26.9	9	19.0		13 44 62	36.2 29.6 17.8	6	4	.008 *	.845

Twelve patients did not meet any of the NCCN criteria.

B: Breast, O: ovarian, C: Cancer C, N: Number of patients, *: Significant.

Table 4. Sensitivity, specificity, and positive and negative predictive values for BRCAPRO and BOADICEA models in the prediction of BRCA1 and BRCA2 mutations.

	Sensitivity			Specificity			PPV			NPV
	BRCAPRO model	BOADICEA Model	Either BOADICEA or BRCAPRO model	BRCAPRO model	BOADICEA model	Either BOADICEA or BRCAPRO model	BRCAPRO model	BOADICEA model	Either BOADICEA or BRCAPRO model	BRCAPRO model	BOADICEA model	Either BOADICEA or BRCAPRO model
	BRCA1/or BRCA2			BRCA1/ or BRCA2			BRCA1/ or BRCA2			BRCA1 / orBRCA2
BC at age ≤ 50 years and ≥ 1 BC/OC close relative at age ≤ 50 years	0.790	0.80	0.78	0.70	0.83	.330	74%	83%	55%	76%	80%	58.5%
Patients with relatives with male BC	0.32	0.32	0.89	1	0.93	390.	%100	54%	61%	63%	76%	77%
Triple negative BC	0.17	0.20	0.75	0.59	0.90	0.22	31%	68%	51%	40%	51%	45%

PPV: Positive predictive value, NPV: Negative predictive value NPV, BC: breast cancer, OC: ovarian cancer.

In total, 62 patients were classified into four molecular subtypes of tumors (Table 2). More than half of the patients had luminal B [39/62 (62.9%)], followed by luminal A [11/62 (17.7%)], triple-negative BC [8/62 (13%)], and HER2+ [4/62 (6.5%)]. Half of the BC patients aged ≤50 years had luminal B subtype (51.6%), whereas 11.3% had luminal A, 9.7% were triple-negative, and 4.8% were HER2 positive. Premenopausal and postmenopausal patients aged ≤50 years with the luminal A subtype represented 4.8% and 6.5%, respectively, while luminal B patients represented 38.7% and 12.9%.

From the BRCAPRO and BOADICEA model scores based on the information provided by the patients, the highest prediction rates were 90.8% for BRCA2 and 75.7% for BRCA1. The patient with the highest BRCA2 score had a first-degree relative and a second-degree relative with a history of male breast cancer and was classified as a luminal B subtype. On the other hand, the highest score for BRCA1 was for a patient with both breast and ovarian cancer. The patient was classified as luminal A.

The mean carrier probability for all mutations in BRCAPRO was 8.1% for BRCA1 and 18.7% for BRCA2. For BOADICEA, it was 4.8% for BRCA1 and 12.3% for BRCA2. Using a cutoff of ≥10%, BOADICEA identified 5/62 (8%) of the patients as having a high risk of BRCA1 mutations and 18/62 (29%) as high risk for BRCA2 mutations, whereas BRCAPRO identified 17/62 (27.4%) for BRCA1 and 26/62 (41.9%) for BRCA2. Consequently, 32/62 (51.6%) of the patients were identified as high risk by BOADICEA and/or BRCAPRO and therefore eligible for genetic testing. Four individuals had a high risk of BRCA1 in both models, 13 had a high risk in BRCAPRO only, and 1 had a high risk in BOADICEA only. For BRCA2, 15 patients were positive in both models at the same threshold: 11 were positive in BRCAPRO, and 3 were positive in BOADICEA.

The average ages (mean ± SD) of patients classified as high risk for BRCA1 mutations by BRCAPRO and BOADICEA were, respectively, 42.5 ± 9.5 years and 43.6 ± 14.8 years. For both models together, it was 42.7 ± 9.2 years. The mean age of patients identified as having a low risk of BRCA1 mutations by these models was 45.1 ± 8.9 years (p = 0.7). On the other hand, the average age of patients with a high risk of BRCA2 mutation by BRCAPRO, BOADICEA, both models and low-risk BRCA2 mutations were 42.12 ± 7.8 years, 43.5 ± 9.6, 43.6 ± 8.4, and 45.8 ± 9.2, respectively (p = 0.8).

Table 3 shows the results for 50/62 (80.6%) patients who met at least one NCCN criterion. Using logistic regression, the NCCN criterion “BC patient at age ≤ 50 with one or more BC close blood relatives” was a statistically significant predictor of patients identified as high risk for BRCA1 and BRCA2 mutations by either BRCAPRO or BOADICEA score (p = 0.011 and p = 0.005, respectively), with sensitivity of 0.78, specificity of 0.33, and area under the ROC curve of 0.72 (Table 4, Figure 1). Furthermore, based on BRCAPRO and BOADICEA scores of high risk BRCA1 or BRCA2 mutation (≥10%) as the reference standard (found in 51.6% of our cohort) to determine the PPV and NPV for each criterion, the BC patient at age ≤50 with one or more BC close blood relative criterion identified patients with a PPV of 55% and a NPV of 58.5%. The criterion of BC patient with male relative with BC was statistically significant in patients at high risk of BRCA2 mutations (p = 0.023), with a sensitivity of 0.89, specificity of 0.39, and area under the ROC curve of 0.75 (Table 4, Figure 2). PPV was 61% and NPV was 77%. Triple-negative BC patients diagnosed at age ≤60 years were also at a statistically significant high risk of BRCA1 mutations (p = 0.008), with a sensitivity of 0.75, specificity of 0.22 and area under the ROC curve of 0.76 (Figure 3). PPV was 51% and NPV 45%.

Figure 1. ROC curve for NCCN criterion of breast cancer at age ≤ 50 years and one or more close blood relatives with breast or ovarian cancer at age ≤ 50 years as a predictor of genetic test eligibility.

Figure 2. ROC curve for NCCN criterion: breast cancer patients with male relatives with BC as a predictor of genetic test eligibility.

Figure 3. ROC curve for NCCN criterion: Triple-negative (ER, PR, HER2-) breast cancer diagnosed at age ≤ 60 years as a predictor of genetic test eligibility.

4. Discussion

Despite recent efforts by Libyan health authorities to develop nationwide breast cancer screening programs [31], the unavailability of genetic diagnostics for people with familial breast and ovarian cancer remains a concern. The situation of BRCA mutations in Libyan women has not been investigated. This is the first study to use BRCAPRO and BOADICEA scoring (≥10%) to predict BRCA1/2 gene mutations in women with BC who have a strong family history of BC and to evaluate the efficacy of NCCN criteria.

Based on the BRCAPRO and BOADICEA scoring systems, our calculated frequencies of patients at high risk of being carriers of BRCA1 and BRCA2 mutations were 29% and 46.8% for the respective scoring systems. If these high-risk patients were to test positive for BRCA1/2 mutations, this estimate would be consistent with studies showing that BRCA2 mutations are more common than BRCA1 mutations in the Arab region [32], though this pattern is not observed in most other populations [33]. We also found that the occurrence of both breast and ovarian cancer in one patient was related to a high risk of a BRCA1 mutation, as reported in Sweden [34]. This woman was 44 years old at the time of diagnosis and had the highest probability of BRCA1 mutation in the BODICEA model.

We employed BRCAPRO and BOADICEA because they are the most reliable in predicting mutant carrier probability when compared to other scoring models [35]. BRCAPRO predicted that 27.4% of the patients had BRCA1 mutations and 41.9% had BRCA2 mutations, which are higher than the percentages predicted by BOADICEA (8% for BRCA1 and 29% for BRCA2). The BRCAPRO model is similar to BOADICEA in that BRCA1 and BRCA2 are modeled independently. The differences between them can be explained in part by their use of different mutation rates and allele frequencies. In particular, in BRCAPRO, the mutations are assumed to be more common in BRCA1 than in BRCA2, but BOADICEA finds that BRCA1 and BRCA2 mutations have similar population frequencies, though higher frequency of BRCA2 has also been reported [18]. The benefit of using two predictive models instead of one is that we were able to identify more high-risk patients who needed genetic testing, and based on score results from either or both scoring systems, we recorded 13 more BRCA1 and 11 more BRCA2 high-risk patients compared to using either system alone.

One of the highest risk categories, according to the NCCN, includes patients with BC at the age of ≤50 years and having first- or second-degree relatives with BC at the age of ≤50 years. The average age of these patients was 44.8 years, which is consistent with a previous study in Libya that showed a higher frequency of BC among those ≤50 years old [3,36–38]. Luminal B subtype (61.3%) was more prevalent than luminal A (17.7%). This is in agreement with studies in Tunisia [39], Morocco [40], Saudi Arabia [41] and Italy [42] but differs from studies in other North African countries, including Egypt, Tunisia, and Algeria [43–45]. The difference could be explained by the heterogeneity of BC in different cohorts drawn from different countries. Furthermore, the 13% triple-negative breast tumors in our study are similar to the percentages found in Tunisia and Morocco [39,40], and the 6.5% for the HER2-positive subtype is in agreement with studies from Algeria, Europe, and the USA [46].

Several studies refer to a link between specific molecular subtypes and BRCA1/2 mutation status. BRCA1/2 mutation carriers are more likely to have triple-negative BC. This association is strongest in BRCA1-related BC, and most BRCA2 BCs belong to the luminal B subtype [47]. Interestingly, our study shows that 75% of the triple-negative BC patients had scores ≥10 of carrier BRCA1 mutation, and 55.3% of the luminal B subtype had BRCA2 scores ≥10. These findings need confirmation by full gene sequence analyzes of BRCA1 and BRCA2 in the same patients.

The overall mutation prevalence in patients identified as high risk by BRCAPRO and BOADICEA was lower than expected based on the NCCN criteria. The NCCN criteria that correlated best with the scoring systems for identifying high-risk patients were as follows: (i) BC patients at age ≤50 years and having one or more close blood relatives with BC at age ≤50 years, (ii) BC patients having male relatives with BC, and (iii) patients with triple-negative BC (ER, PR, and HER2) diagnosed at age ≤60 years. For these, sensitivity was high (78%, 89%, and 75%, respectively) and specificity was low (33%, 39%, and 22%, respectively). This could result in a large number of patients with a low likelihood of having a BRCA mutation being referred for a genetic test. Though the p value of BC patients with BRCA2 at age ≤45 years was 0.032, it had a poor area under the ROC curve (0.4), indicating that the criterion may not differentiate between high and low risk BRCA mutation carriers, especially because it depends on patient age alone. The other criteria (double primary BC and one or more relatives with prostate or pancreatic cancer) were not significant, possibly due to the small sample size.

Our study has some limitations. We hypothesized that a BRCAPRO or BOADICEA cutoff of ≥10% would be enough to predict the presence of a BRCA mutation in patients who require a genetic test with acceptable false-negative rates. We could not compensate for the fact that none of our patients had a genetic test. However, this study aimed to make a pretest prediction of BRCA1 and BRCA2 carrier mutations. We also had only one patient with double primary BC, one with double primary breast and ovarian cancer, four BC patients with relatives who had prostate or pancreatic cancer, and eight triple-negative (ER, PR, and Her2) BC patients at the age ≤60. As a result, increasing the number of patients who meet the rarer criteria will improve the statistical analysis of that criterion.

5. Conclusion

Our study sheds light on the complexities of the NCCN criteria for genetic testing decision-making in the context of BRCA1 and BRCA2 mutations. While we did not intend to provide a definitive judgment on the validity of these criteria, our findings indicate areas where they demonstrate strengths and areas where they may benefit from further refinement. The NCCN criteria serve as a valuable guideline for genetic testing, providing a critical framework for healthcare providers. However, certain elements within the NCCN guidelines displayed varying levels of significance when compared to the models examined in our study. This suggests that some items within the NCCN criteria may be more predictive and informative than others in the context of BRCA1 and BRCA2 mutation testing.

Furthermore, our study also indicates that the models we examined show promise in enhancing the decision-making process for genetic testing. While we cannot affirm that the models are universally superior to the NCCN criteria, our results suggest that there may be situations or contexts in which these models offer advantages in terms of specificity, sensitivity, or other relevant metrics. Enhancing risk assessment could lead to improved treatment outcomes by using more accurate selection of patients for testing and therapy and a proactive approach to managing the risk of breast cancer of patients and their families.

In light of these findings, we recommend that further research be conducted to identify specific areas where the NCCN criteria could be refined to enhance their performance, as well as to explore the circumstances in which alternative models might prove advantageous. Ultimately, a comprehensive and dynamic approach to genetic testing decision-making is essential, considering the evolving landscape of genomic medicine and the potential for improved patient care. This study’s results can be validated by performing germline BRCA1/2 testing based on the BRCAPRO and BOADICEA models.

Availability of data and materials

All data generated or analyzed during this study are included in supplementary information files (S1, S2, and S3).

Author contributions

Conceptualizat ion: Eanas Elmaihub, Inas Alhudiri

Data curation: Eanas Elmaihub, Fakria Elfagi

Formal analysis: Eanas Elmaihub

Investigation: Eanas Elmaihub, Fakria Elfagi

Methodology: Eanas Elmaihub, Inas Alhudiri

Supervision: Elham Hassen, Adam Elzagheid

Writing – original draft: Eanas Elmaihub

Writing – review and editing: Elham Hassen, Inas Alhudiri

Ethical considerations

The study was approved by the Research Ethics Committee of the National Cancer Institute (NCI), Sabratha, Libya. Informed consent was obtained from each patient before recruitment in the study, which had been approved by the Research Ethics Committee of the National Cancer Institute (NCI), Sabratha, Libya. This study was conducted in accordance with the ethical principles of the Helsinki Declaration.

Acknowledgments

We appreciate the help received from the Sabratha Research Unit of the National Cancer Institute and the Oncology Department.

Disclosure statement

No potential conflict of interest was reported by the authors.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/20905068.2024.2341368

Additional information

Notes on contributors

Eanas Saleh Elmaihub

Eanas Saleh Elmaihub is an academic teacher and biomedical researcher interested in molecular biology, biotechnology, and genetic oncology research. She received her Bachelor of Zoology from Benghazi University (Faculty of Sciences) in Libya and her Master’s degree in Biomedical and Biology from the Libyan Academy, Tripoli, Libya. She is currently in her fifth year as a Ph.D. student in molecular biology and human genetics at Monastir University, Tunis.

Eanas also serves as the Head of the Molecular biology and biochemistry department at Sciences College, Sabratha University, Libya. In addition, she is a member of the research team of the Molecular Diagnostic Unit at the National Cancer Institute (NCI), Sabratha, Libya. Eanas has also held several other positions, including a member of the Scientific Committee at the Faculty of Sciences at Sabratha University and a team member for combating COVID-19 at NCI. Elmaihub’s research interests include familial breast, ovary, and colon cancer, the epidemiology of breast and ovary cancer, gene variants, and mutations among the Libyan population, and molecular diagnostics of parasites and bacteria. She has been participating with several organizations, like AGYA, on rare disease research and with WHO on different subjects, like biological risk management. She has supervised numerous bachelor’s students. For more information, please see the following links:

Google Scholar: https://scholar.google.com/citations?hl=ar&user=h6Rz4REAAAAJ

Research Gate: https://www.researchgate.net/profile/Eanas-Elmaihub

Inas Alhudiri

Dr. Inas Alhudiri is a medical professional and experienced researcher in genetics and biotechnology. She holds an MB ChB from Sabha University and an MRCS from the Royal College of Physicians and Surgeons of Glasgow. She completed her PhD from the University of Nottingham in 2012. Currently, Dr. Alhudiri serves as the Head of the Genetic Engineering Department and the Libyan Biobanks project at the Libyan Biotechnology Research Centre in Tripoli. She has held several other key positions, including Chairman of the Technical Committee for Combating COVID-19 and Director of the Vaccination Unit. Dr. Alhudiri’s research interests include assessing genomic diversity, breast and colon cancer risk assessment, and investigating breast cancer gene variants and mutations. She has participated in various collaborative research projects and is a Professor of Molecular Biology and Medical Genetics at the School of Life Sciences at the Libyan Academy. She has supervised several master’s students during their research year and is currently supervising four students in their final stage of the project. For more information, see the following links: Google scholar: https://scholar.google.com.ly/citations?user=WK68w1EAAAAJ&hl=en ResearchGate: https://www.researchgate.net/profile/Inas-Alhudiri

Adam Elzagheid

Dr. Adam Elzagheid is a professor at Libyan Biocenology Research Centre (LBTRC), acting as General Director of LBTRC, Tripoli, Libya. Dr. Elzagheid worked as a Dean of Faulty of Medicine, Benghazi University, Benghazi, Libya, Head of Department of Pathology, Faculty of Medicine, University of Benghazi, Post-doctoral fellow and Research Associate at University of Turku, Faculty of Medicine, Oncology and Pathology Departments, Turku, Finland. Over the last twenty years at Turku University, Department of Pathology and oncology (Turku, Finland), University of Benghazi, Department of Pathology (Benghazi, Libya), (LBTRC) Libyan Biotechnology Research Centre (Tripoli, Libya), Dr. Elzagheid gained excellent teaching, research, administrative and managerial experience, and this has been documented by interesting research paper. Research & development experience in the prognostication of tumor biomarkers in solid tumors (such as; breast, colorectal, and prostate tumors). Professional experience in curriculum development, articulations, accreditation, professional certification, and Training. Teaching experience in general and systemic pathology, Research Methodology, molecular pathology, Lab techniques. For more information, following links: https://scholar.google.com/citations?user=EQQpGXAAAAAJ&hl=en https://www.ncbi.nlm.nih.gov/pubmed/?term=Elzagheid+A https://www.researchgate.net/profile/Adem_Elzagheid

Fakria Elfagi

Dr. Fakria Elfagi, a highly regarded medical oncologist specializing in solid tumors, boasts a career dedicated to oncology. After her MBBCh (1990) and internship (1990-1991), Dr. Elfagi honed her skills in pediatrics (1991-1999) and hematology (1999-2001) before specializing in medical oncology (2001-2006) at the University of Catanzaro (Magna Grecia) in Italy. Returning to Libya, she served as a medical oncology specialist (2006-current 2024) at the Sabrata National Cancer Institute, leading the department (2011-2020). Dr. Elfagi’s passion extends to teaching internal medicine at Zawia Medical Faculty and contributing to the National Cancer Control Program. Her active participation in scientific meetings underscores her commitment to continuous learning and knowledge sharing.

Elham Hassen

Dr. Elham Hassen is a professor at the Institut Supérieur de Biotechnologie de Monastir and the head of the Immuno-oncology laboratory at Tunisia’s Medical College Monastir University. She had extensive research and academic teaching experience with postgraduate students, and she supervised master’s and doctoral students. Elham documented a lot of interesting studies on the genetics of breast and nasopharyngeal cancer. She was also very knowledgeable about single nucleotide polymorphism analysis. For more information, please see the following links: https://pubmed.ncbi.nlm.nih.gov/?term=Hassen+E&cauthor_id=27852262 https://www.researchgate.net/profile/Elham-Hassen