Evaluation of tumor-infiltrating lymphocytes across breast cancer stages at the University of Gondar Comprehensive Specialized Hospital, Northwest Ethiopia Authors: Emiyamrew Getnet, Gashaw Adane, Ephrem Awoke Shiferaw, Tadele Ayelegn Nibret, Gebremariam Maru Yemru, Minyahl Kassye Mengstu, Abebe Birhanu, Afewerk Habtamu, Habtu Debash, Tadelo Wondmagegn, Mulualem Lemma Published: 2026 ABSTRACT BackgroundTumor-infiltrating lymphocytes (TILs) are recognized as prognostic biomarkers. However, inconsistent TIL scores across breast cancer stages, and data are scarce for Ethiopian patients. Thus, this study evaluated TIL scores across breast cancer stages in Northwest Ethiopia.MethodsA cross-sectional study was conducted from 20 June 2024 to 19 June 2025. Using consecutive sampling, 62 breast cancer patients were enrolled, and their socio-demographic and clinicopathological data were collected using questionnaires and chart reviews. Formalin-fixed, paraffin-embedded tissue sections were stained with hematoxylin and eosin, and stromal TILs were evaluated according to the International TIL Working Group guidelines. Data were analyzed using SPSS version 27. Median TILs across clinicopathological variables were assessed using Kruskal-Wallis and Mann-Whitney U tests. Spearman correlation assessed the association of TIL scores and clinicopathological variables. P ≤ 0.05 was considered statistically significant.ResultsLow TIL scores were observed in 76.5% of stage II cases, whereas high TIL scores were observed in 54.5% of stage IV cases. Median TIL percentages in stage IV (48%) and III (33%) were significantly higher than in stage II (7%), and in N2–N3 (33%) than in N0–N1 (9%) (p < 0.001). Moreover, the median TIL percentage in grade III (48%) was higher than in grade I (9%) (p = 0.028). TILs were positively correlated with TNM stage, lymph node stage, and grade.ConclusionsA positive correlation was found between TIL score and TNM stage. Intermediate and high TIL scores were common in advanced stages, whereas low TIL scores were prevalent in early stages. The findings suggest that advanced tumors may harbor higher mutation burdens and immunogenicity. Future research should explore the molecular subtypes of immune profiles. Background Breast cancer (BC) is the most commonly diagnosed cancer worldwide, with about 2.3 million new cases and 685,000 deaths per year [1]. Historically, incidence and mortality rates have been higher in developed countries than in developing countries [1–3]. However, both incidence and death rates have been increasing in sub-Saharan Africa [4]. In Ethiopia, BC remains a major public health concern, with an estimated 16,133 new cases and 9,061 deaths annually [1], accounting for 22.3% of all cancer-related deaths [5]. Most patients are diagnosed at advanced stages at the University of Gondar Comprehensive Specialized Hospital (UoG-CSH) [6]. Antigen-presenting cells recognize tumor antigens and migrate to lymph nodes (LNs) to activate lymphocytes, which then infiltrate the tumor microenvironment (TME) as TILs, including T cells, B cells, and natural killer (NK) cells [7–9]. Based on their location, TILs are divided into stromal and intratumoral types. Stromal TILs are lymphocytes distributed throughout the tumor stroma, whereas intratumoral TILs are located within the tumor nests and have direct contact with cancer cells [10, 11]. Stromal TILs are considered more reproducible than intratumoral TILs due to their higher frequency and ease of detection using hematoxylin and eosin (H&E) staining and without the need for immunohistochemistry/fluorescence techniques [11, 12]. Tumor-infiltrating lymphocytes are theoretically correlated with BC outcomes due to their role in targeting cancer cells and predicting a favorable prognosis [10, 13, 14]. However, beyond their antitumor immunity, TILs can also promote tumor aggressiveness by facilitating local tissue invasion and metastasis [15–18]. The composition of TILs varies by BC subtypes [19], disease stages [20, 21], and patient races [22–25]. The American Joint Committee on Cancer (AJCC) system classifies BC into early stages (0–II) and advanced stages (III–IV) based on tumor size (T), LN involvement, and distant organ metastasis (M) [26]. Moreover, the relative densities of TILs have shown inconsistent correlations with BC stages. Studies have shown that high stromal TIL scores are positively correlated with advanced stages and positive LN status [21]. In contrast, other studies found that low TIL scores correlated with advanced stages and positive LN status [27]. Some studies found no statistically significant correlation between TILs and T size or LN status [28]. Assessing TILs is essential for determining patients’ therapeutic responses, as TIL composition affects response to therapy. Tumors with high TILs often express higher levels of programmed cell death ligand 1 (PD-L1) [29, 30] and are more likely to respond to immunotherapy, whereas low-TIL scores may be responsive to chemotherapy. Thus, this study aimed to assess TILs across BC stages in Northwest Ethiopia. This study provides a foundation for future immunological and molecular investigations. Materials and methods Study design, period, and setting A hospital-based cross-sectional study was conducted from 20 June 2024 to 19 June 2025 at the Oncology Treatment Center of UoG-CSH in Northwest Ethiopia, 750 km from the capital, Addis Ababa. At the center, cancer diagnosis primarily depends on clinicopathological examinations. Imaging techniques such as computed tomography (CT), ultrasound, and bone scintigraphy are used to assess distant metastasis. The center has separate pediatric and adult oncology units with more than 40 beds and treats nearly 3,000 cancer cases annually [31]. Study population and selection criteria The source population comprised all adult female patients with histopathologically confirmed BC at the UoG-CSH pathology laboratory. The study included all adult women diagnosed with invasive BC based on H&E-stained sections of formalin-fixed paraffin-embedded tissue before receiving neoadjuvant therapy. Over the study period, 75 incisional, excisional, and mastectomy specimens were consecutively obtained at the pathology laboratory before neoadjuvant therapy. Mastectomy specimens were sometimes obtained from stage IV patients with limited distant metastasis to reduce tumor burden and obtain sufficient tissue for definitive diagnosis [32]. To avoid confounding effects on TILs, patients with the following comorbidities were excluded: Coexisting malignancies may modify systemic immunity and thereby affect TIL scores [33]; diabetes mellitus may reduce TIL scores [34, 35]; hypertension managed by calcium channel blockers can decrease TIL scores [36]; and infectious diseases, including HIV, alter immune cell activation and tumor infiltration [37]. Patients with prior core biopsies were excluded because biopsy-induced tissue inflammation can artificially alter TIL scores [38]. Consequently, from the initial 75 BC, 13 were excluded: 6 with hypertension, 3 with diabetes mellitus, 1 HIV-positive, 1 had cervical cancer, and 2 had prior core biopsies. Finally, 62 BC patients were included, and their socio-demographic and clinical data were collected using interview-administered questionnaires and medical chart reviews (Fig. 1). Fig. 1 Breast cancer patient selections and TILs evaluation workflows guidelines [11] Breast cancer patient selections and TILs evaluation workflows guidelines [11] Tissue processing and histopathological evaluation Mastectomy, incisional, and excisional biopsy specimens were sent to the pathology laboratory fully submerged in a container filled with 10% neutral buffered formalin. Upon arrival, the specimens underwent standard gross examination, followed by tissue processing: dehydration in ethanol, clearing in xylene, and paraffin embedding using a Leica TP 1020 tissue processor. Then, 4–5 micrometer-thick tissue sections were prepared and stained with H&E. Subsequently, all stained sections were examined for LN status, histologic grades, and histologic types [26]. Evaluation and scoring of tumor-infiltrating lymphocytes Microscopic evaluation of TILs on H&E-stained tissue sections was performed according to the International TIL Working Group guidelines [11]. The evaluation considered only mononuclear cells, particularly lymphocytes and plasma cells, located between tumor nests and within the invasive tumor borders. Pathologists assessed slides at sufficiently high magnification to accurately distinguish lymphocytes from neutrophils, eosinophils, basophils, and histiocytes. Lymphocytes located outside the tumor borders, adjacent to ductal carcinoma in situ, or within normal lobules, as well as in regions with crush artifacts, extensive central necrosis, fibrosis, or prior biopsy sites, were excluded from the assessments. The percentage of the total stromal area occupied by lymphocytes and plasma cells was determined across multiple randomly selected microscopic fields, without hotspot-focused assessment. Finally, TILs were scored as continuous percentages and subsequently categorized into three predefined groups: low (≤ 10% TILs), intermediate (11% ≤ TILs ≤ 40%), and high (≥ 41%) at 200-400x magnification, as the working group guidelines [11]. Low TILs are characterized by the absence of, or minimal, lymphocyte infiltrations, resulting in the visible portions of the stroma appearing lighter. In contrast, intermediate TILs are more abundant than low TILs. However, with high TILs, the lymphocyte infiltrations were dense, making the stromal area appear darker (Fig. 2). In cases where TILs were heterogeneously distributed, multiple areas are assessed, and reference stromal TIL images are used [39, 40]. Fig. 2Histopathological appearance of TIL score of H&E-stained slides (40x) of BC cases: (A) low TILs; (B & C) intermediate TIL scores; (D) high TIL scores; and (E) high TIL scores at 100x for validation at the department of pathology of UOG-CSH, Ethiopia Histopathological appearance of TIL score of H&E-stained slides (40x) of BC cases: (A) low TILs; (B & C) intermediate TIL scores; (D) high TIL scores; and (E) high TIL scores at 100x for validation at the department of pathology of UOG-CSH, Ethiopia Quality assurance We strictly followed all standard operating procedures (SOPs) for tissue processing and ensured optimal H&E staining quality. We performed regular maintenance and replaced tissue processing reagents every two (2) weeks with fresh reagents. Stromal TIL evaluations were performed by four pathologists who were blinded to clinical information, thereby minimizing potential bias. In addition, pathologists used reference percentage TILs to ensure consistency and reproducibility. Data analysis Collected data were entered and analyzed using Statistical Package for Social Sciences (SPSS) version 27 (IBM Corporation, Armonk, NY, USA). Categorical variables were summarized as frequencies and percentages. TILs were analyzed as a categorical variable and as a continuous variable using the median with interquartile range, since the data were not normally distributed (Shapiro–Wilk test, p < 0.05). Fleiss kappa (κ) was used to assess inter-observer agreement among pathologists evaluating TIL scores [41]. The levels of agreement are κ < 0.00 (poor); κ = 0.00-0.20 (slight); κ = 0.21–0.40 (fair); κ = 0.41–0.60 (moderate); κ = 0.61–0.80 (substantial); and κ = 0.81-1.00 (perfect) [42]. Disagreement cases between at least three of the four pathologists were reviewed together by all four pathologists using a multi-headed microscope and discussed until consensus scores were reached [43]. If consensus could not be reached, the median scores of the four pathologists were used. To assess median TIL percentage differences across TNM stages and histological grades, the Kruskal-Wallis test followed by Dunn’s post hoc pairwise comparisons with Bonferroni correction was used. The Mann-Whitney U test was used to assess TIL differences between two groups. Spearman correlation coefficient (ρ) was used to determine the correlation between TIL scores (ordinal variables) and clinicopathological parameters. The strength of correlation was classified as: no correlation (0 ≤ |ρ| < 0.19), weak (0.2 ≤ |ρ| < 0.39), moderate (0.4 ≤ |ρ| < 0.69), and strong (|ρ| ≥ 0.7) [44]. A p-value of ≤ 0.05 was considered statistically significant. Results Socio-demographic and behavioral characteristics of breast cancer patients This study analyzed data from 62 BC patients, with a mean age of 42.17 years; most participants (69.4%) were under 50 years of age. None of the participants reported smoking or alcohol consumption, and 98.39% had no family history of BC. 74.2% of the patients were contraceptive users, while 25.8% did not use any contraceptive method (Table 1). Table 1Socio-demographic characteristics of breast cancer patients at UoG-CSH, Northwest Ethiopia, 2025VariablesCategoriesFrequencyPercentage (%)Age (years)< 504369.4≥ 501930.6Age at parity< 201422.58≥ 203556.45Nulliparous1320.97ResidenceRural3658.1Urban2641.9Family history of BCYes11.61No6198.39Contraceptive usageNo-user1625.8Oral pills711.3Syringe injection2540.3Implant914.5Injection & implant58.1Of 62 BC patients, 49 had given birth, and 13 were nulliparous (never having given birth) Table 1Socio-demographic characteristics of breast cancer patients at UoG-CSH, Northwest Ethiopia, 2025VariablesCategoriesFrequencyPercentage (%)Age (years)< 504369.4≥ 501930.6Age at parity< 201422.58≥ 203556.45Nulliparous1320.97ResidenceRural3658.1Urban2641.9Family history of BCYes11.61No6198.39Contraceptive usageNo-user1625.8Oral pills711.3Syringe injection2540.3Implant914.5Injection & implant58.1Of 62 BC patients, 49 had given birth, and 13 were nulliparous (never having given birth) Socio-demographic characteristics of breast cancer patients at UoG-CSH, Northwest Ethiopia, 2025 Of 62 BC patients, 49 had given birth, and 13 were nulliparous (never having given birth) Clinicopathological characteristics of pre-treatment breast cancer patients According to the tumor characteristics of the study participants, a right-sided tumor was the most frequently found (54.83%), and invasive ductal carcinoma was the most prevalent histologic type (88.7%). Most patients had grade III (50.0%) and grade II (40.0%). Regarding T stages, T2 was the most common (45.2%), followed by T3 (43.5%), while T1 and T4 accounted for 8.1% and 3.2%, respectively. Based on LN status, N2 was the most frequent (41.94%), followed by N1 (25.81%), N0 (17.74%), and N3 (14.51%). In terms of overall TNM stages, two (2) patients were in stage I, 17 in stage II, 21 in stage III, and 22 in stage IV (Table 2). Table 2Clinicopathological characteristics of breast cancer patients, Ethiopia, 2025VariablesCategoriesFrequencyPercentage (%)Types of specimensMastectomy4267.7Incisional/excisional biopsies2032.3Tumor lateralityRight3454.83Left2743.55Bilateral11.62Histological gradeI (well differentiated)812.9II (differentiated)2438.71III (poorly differentiated)3048.39Histopathological typeInvasive ductal carcinoma5588.7Invasive lobular carcinoma58.1Papillary carcinoma11.60Mucinous carcinoma11.60T sizeT1(< 2 cm)58.1T2 (2–5 cm)2845.2T3 (≥ 5 cm)2743.5T4 (spread to chest or skin)23.2LN statusN0 (no LN spread)1117.74N1 (1–3 LN spread)1625.81N2 (4–9 LN spread)2641.94N3 (≥ 10 LN spread)914.51Distant organ MetastasisM0 (no metastasis)4064.52M1(metastasized)2235.48TNM stagesI23.2II1727.4III2133.9IV2235.5Stage I (T1N0); stage II (T0N1, T1N1, T2N0, T2N1, & T3N0), and stage III (T0N2, T1N2, T2N2, T3N1, T3N2, T4N0, T4N1, T4N2, & any TN3), and stage IV (any T, any N, and M1) Table 2Clinicopathological characteristics of breast cancer patients, Ethiopia, 2025VariablesCategoriesFrequencyPercentage (%)Types of specimensMastectomy4267.7Incisional/excisional biopsies2032.3Tumor lateralityRight3454.83Left2743.55Bilateral11.62Histological gradeI (well differentiated)812.9II (differentiated)2438.71III (poorly differentiated)3048.39Histopathological typeInvasive ductal carcinoma5588.7Invasive lobular carcinoma58.1Papillary carcinoma11.60Mucinous carcinoma11.60T sizeT1(< 2 cm)58.1T2 (2–5 cm)2845.2T3 (≥ 5 cm)2743.5T4 (spread to chest or skin)23.2LN statusN0 (no LN spread)1117.74N1 (1–3 LN spread)1625.81N2 (4–9 LN spread)2641.94N3 (≥ 10 LN spread)914.51Distant organ MetastasisM0 (no metastasis)4064.52M1(metastasized)2235.48TNM stagesI23.2II1727.4III2133.9IV2235.5Stage I (T1N0); stage II (T0N1, T1N1, T2N0, T2N1, & T3N0), and stage III (T0N2, T1N2, T2N2, T3N1, T3N2, T4N0, T4N1, T4N2, & any TN3), and stage IV (any T, any N, and M1) Clinicopathological characteristics of breast cancer patients, Ethiopia, 2025 Stage I (T1N0); stage II (T0N1, T1N1, T2N0, T2N1, & T3N0), and stage III (T0N2, T1N2, T2N2, T3N1, T3N2, T4N0, T4N1, T4N2, & any TN3), and stage IV (any T, any N, and M1) Stromal TIL scores and correlations with breast cancer TNM stages The assessments of TIL by four pathologists showed substantial agreement, with Fleiss kappa values of 0.75. Based on this reliability, the correlation between TILs and stages was analyzed. Regardless of lymphocyte subsets, low TIL scores were frequently observed in the early stages (I and II). In stage II, low TIL scores accounted for 76.5% of cases, whereas high TIL scores were observed in only 5.9% of patients. Intermediate TIL scores were the most frequently observed in stage III, comprising 57.1% of cases. In contrast, stage IV exhibited a predominance of high TIL scores (54.5%), with 13.6% of low TIL scores. A moderate, significant correlation was found between TIL scores and overall TNM stages (ρ = 0.534, p < 0.001). Similarly, a significant positive correlation was observed between LN stages and TIL scores (ρ = 0.434, p < 0.001). Low TIL scores were mainly found in N0 (72.73%) and N1 (43.75%), while high TIL scores were primarily found in N3 (55.55%). TIL scores also positively correlated with distant organ metastasis (ρ = 0.378, p = 0.002), histologic type (ρ = 0.273, p = 0.032), and grade (ρ = 0.385, p = 0.002). In contrast, no statistically significant correlation was found between TIL scores and T sizes (p = 0.301). Low TIL scores were predominantly found in T1, whereas T2 and T3 showed an even distribution of scores among the three TIL categories (Table 3). Table 3Correlations of TILs with clinicopathological variables of BC, Ethiopia, 2025VariablesStromal TILs scoreΡP-valueLow (%)Intermediate (%)High (%)TNM stages I2(100.0)0(0.0)0(0.0) 0.534 < 0.001 II13(76.5)3(17.6)1(5.9) III2(9.53)12(57.14)7(33.33) IV3(13.64)7(31.82)12(54.54)Histologic grades Grade I6(75.0)2(25.0)0(0) 0.385 0.002 Grade II6(25.0)14(58.3)4(16.7) Grade III8(26.7)6(20.0)16(53.3)Histologic types Ductal carcinoma16 (29.1%)19 (34.5%)20 (36.4%) 0.273 0.032 Non-ductal types4 (57.1%)3 (42.9%)0 (0.0%)T stages T14(80.0)1(20.0)0(0.0)0.1330.301 T28(28.57)9(32.14)11(39.29) T38(29.63)11(40.74)8(29.63) T40(0.0)1(50.0)1(50.0)LN stages N08(72.73)3(27.27)0(0.0) 0.434 < 0.001 N17(43.75)3(18.75)6(37.5) N24(14.8)13(51.9)9(33.3) N31(11.11)3(33.33)5(55.55)Distant organ Metastasis M017(42.5)15(37.5)8(20.0) 0.378 0.002 M13(13.64)7(31.82)12(54.54) Table 3Correlations of TILs with clinicopathological variables of BC, Ethiopia, 2025VariablesStromal TILs scoreΡP-valueLow (%)Intermediate (%)High (%)TNM stages I2(100.0)0(0.0)0(0.0) 0.534 < 0.001 II13(76.5)3(17.6)1(5.9) III2(9.53)12(57.14)7(33.33) IV3(13.64)7(31.82)12(54.54)Histologic grades Grade I6(75.0)2(25.0)0(0) 0.385 0.002 Grade II6(25.0)14(58.3)4(16.7) Grade III8(26.7)6(20.0)16(53.3)Histologic types Ductal carcinoma16 (29.1%)19 (34.5%)20 (36.4%) 0.273 0.032 Non-ductal types4 (57.1%)3 (42.9%)0 (0.0%)T stages T14(80.0)1(20.0)0(0.0)0.1330.301 T28(28.57)9(32.14)11(39.29) T38(29.63)11(40.74)8(29.63) T40(0.0)1(50.0)1(50.0)LN stages N08(72.73)3(27.27)0(0.0) 0.434 < 0.001 N17(43.75)3(18.75)6(37.5) N24(14.8)13(51.9)9(33.3) N31(11.11)3(33.33)5(55.55)Distant organ Metastasis M017(42.5)15(37.5)8(20.0) 0.378 0.002 M13(13.64)7(31.82)12(54.54) Correlations of TILs with clinicopathological variables of BC, Ethiopia, 2025 Stromal TILs across clinicopathological variables of breast cancer patients The median TIL percentages in stage IV (48%) and III (33%) were significantly higher than in stage II (7%) (p < 0.001). Similarly, median TIL percentage in N2–N3 (33%) was significantly higher than in N0–N1 (9%) (p < 0.001). The median TIL percentage was significantly higher in grade III (48%) than in grade I (9%) (p = 0.028). Ductal carcinoma also showed a higher median TIL percentage (27%) compared to non-ductal histologic types (8%) (p = 0.039). No statistically significant differences were observed in median TIL percentage by tumor laterality (p = 0.672) and T stage (p = 0.253) (Table 4). Table 4Median values of TILs across clinicopathological variables in BC, Ethiopia, 2025VariablesCategoriesNumber of casesMedian TILs % (IQR)P-valuePairwise comparisonTNM stages bII177 (5-12.5) < 0.001 II vs. III p< 0.001II vs. IV p< 0.001III vs. IV p = 0.736III2133 (22.5–70.5)IV2248 (22.8–70.5)Histologic grades bGrade I89 (7–27) 0.015 GI vs. GII p = 0.624GI vs. GIII p= 0.028GII vs. GIII p = 0.147Grade II2423 (13–33)Grade III3048 (9.8–75.8)T stages aT1 & T23322 (6.5–54.5)0.253T3 & T42927 (10-61.5)Tumor laterality aRight3428 (10–57)0.672Left2725 (8.5–68.5)LN stages aN0 & N1279 (5–35) < 0.001 N2 & N33533 (22–72)Histology types aDuctal carcinoma5527 (10–64) 0.039 Non-ductal types78 (5–23)a Mann-Whitney U testb Kruskal-Wallis testIQR: interquartile range, non-ductal (lobular carcinoma, papillary carcinoma, mucinous carcinoma), vs.: versus. One bilateral case and two stage I cases were excluded from the analysis due to a small sample size Table 4Median values of TILs across clinicopathological variables in BC, Ethiopia, 2025VariablesCategoriesNumber of casesMedian TILs % (IQR)P-valuePairwise comparisonTNM stages bII177 (5-12.5) < 0.001 II vs. III p< 0.001II vs. IV p< 0.001III vs. IV p = 0.736III2133 (22.5–70.5)IV2248 (22.8–70.5)Histologic grades bGrade I89 (7–27) 0.015 GI vs. GII p = 0.624GI vs. GIII p= 0.028GII vs. GIII p = 0.147Grade II2423 (13–33)Grade III3048 (9.8–75.8)T stages aT1 & T23322 (6.5–54.5)0.253T3 & T42927 (10-61.5)Tumor laterality aRight3428 (10–57)0.672Left2725 (8.5–68.5)LN stages aN0 & N1279 (5–35) < 0.001 N2 & N33533 (22–72)Histology types aDuctal carcinoma5527 (10–64) 0.039 Non-ductal types78 (5–23)a Mann-Whitney U testb Kruskal-Wallis testIQR: interquartile range, non-ductal (lobular carcinoma, papillary carcinoma, mucinous carcinoma), vs.: versus. One bilateral case and two stage I cases were excluded from the analysis due to a small sample size Median values of TILs across clinicopathological variables in BC, Ethiopia, 2025 II vs. III p< 0.001 II vs. IV p< 0.001 III vs. IV p = 0.736 GI vs. GII p = 0.624 GI vs. GIII p= 0.028 GII vs. GIII p = 0.147 a Mann-Whitney U test b Kruskal-Wallis test IQR: interquartile range, non-ductal (lobular carcinoma, papillary carcinoma, mucinous carcinoma), vs.: versus. One bilateral case and two stage I cases were excluded from the analysis due to a small sample size Discussion Breast cancer showed different TIL scores, reflecting diverse adaptive immune responses. The densities of TILs depend on the tumor mutational burden and the production of neoantigens. Studies have demonstrated that TILs within the TME play a role in controlling and eliminating tumor cells [45, 46]. However, this classic perspective evolved to recognize that lymphocytes can have pro-tumor and anti-tumor characteristics [16]. In the current study, high TIL scores were primarily observed in stages III–IV, whereas low TIL scores were observed in stages I–II. The median TIL percentages were significantly higher in stage IV (48%) and stage III (33%) than in stage II (7%), and in N2–N3 (33%) than in N0–N1 (9%) (p < 0.001). Similar studies in Brazil [47] and Egypt [48] found that stages III and IV show higher TILs than stage II. A significant positive correlation was observed between TIL score and advanced TNM and LN stages. Similar studies conducted in Serbia [49] and India [21] reported that TIL scores positively correlated with advanced stages. Studies in Brazil [18], Serbia [49], India [21], and Finland [50] also reported that TIL scores significantly increased with advancing LN stages. One possible explanation for this positive correlation is that advanced tumors may have a higher mutational burden, leading to neoantigen production, enhanced immunogenicity, and increased lymphocyte infiltration [21, 49, 51]. Mutational burden is higher in human epidermal growth factor receptor 2 (HER2)-positive and triple-negative BC (TNBC) than in hormone receptor-positive subtypes [52]. Studies in Ethiopia have identified immunogenic TNBC and HER2-positive subtypes [53, 54], which are mostly associated with advanced stages. Furthermore, several hypotheses may explain why TIL scores correlate with advanced stages. The most accepted hypothesis is that certain TIL subsets may secrete growth factors that promote lymphangiogenesis, thereby facilitating cancer spread to LNs [55–58]. Another hypothesis is that TILs produce matrix metalloproteinase 9, which may degrade the extracellular matrix, allowing cancer cells to enter the lymphatic or bloodstream and progress to advanced stages [59, 60]. Additionally, tumors with high TIL scores often express PD-L1 [29, 30] and undergo immune editing, a process in which cancer cells acquire mutations that produce resistant clones that evade immune detection [61, 62]. Moreover, TILs may produce immunosuppressive cytokines such as interleukin-4 (IL-4), IL-10, and transforming growth factor-beta (TGF-β) [63]. In contrast to our findings, an Italian study [27] found that low TIL scores correlated with advanced stages (III & IV) and positive LN status. A study in India [64] also found that low TIL score significantly correlated with LN-positive status. This discrepancy may be influenced by racial TIL distributions [65, 66]. Previous studies have shown higher TIL scores in the TME of Black BC patients than in White patients [22, 67, 68]. Our findings also contrast with a Japanese study [69] that found low T cell infiltration was significantly correlated with advanced stages, and high T cell infiltration with negative LN status. This discrepancy may be due to variations in TIL assessment methods and tumor selection [70]; our study employed standard H&E staining to assess only stromal TILs, while the Japanese study used immunoperoxidase to assess both stromal and intratumoral lymphocytes. A study in Nepal [71] found no significant correlation between TILs and LN status. In the Nepal study, the majority of low TILs were observed in hormone receptor-positive tumors, which may explain the lack of a statistically significant correlation between TILs and LN status [71]. Our study found a significantly higher median TIL percentage in grade III than in grade I (p = 0.028). A similar study in India found that the mean TIL in grade III was higher than in grade I [72]. This finding indicates that TIL densities increase with histologic grade, possibly reflecting greater immunogenicity of high‑grade tumors [73]. A significant positive correlation between TIL score and histologic grade has also been reported in studies conducted in Serbia [49] and Sweden [74]. In contrast, studies from Indonesia [28] and Kenya [75] found no significant correlation between TIL and histologic grade. In addition, our study found the median TIL percentage was higher in invasive ductal carcinoma (27%) than in non-ductal carcinoma (8%) (p = 0.039). A similar study in Belgium found that invasive ductal carcinoma had significantly higher TIL levels than those of lobular carcinoma [76]. Studies in India [64] found a positive correlation between TILs and histologic type. Our study found no significant correlation between TILs and T stages (p = 0.301). This lack of significance may be attributed to small sample sizes, particularly in T1 (n = 5) and T4 (n = 2). A similar non-significant relationship between TILs and T stages was found in Indonesia [28]. Conversely, a study in India [77] showed a positive correlation between TILs and T size, and a study in Nepal [71] reported that low TIL scores were correlated with T3. The Fleiss kappa value of 0.75 indicates substantial inter-observer agreement on TIL assessment among four pathologists. This concordance indicates that approximately 75% of the difference in TIL scoring is attributable to biological differences rather than individual inconsistencies [78]. Previous studies have reported kappa values of fair to moderate concordance [43, 79], reflecting ongoing challenges in reproducibility [80]. Limitations of the study Our study had several limitations. A major limitation is the absence of tumor receptor status data, including estrogen, progesterone, HER2, Ki-67, and intrinsic molecular subtypes, potentially overlooking important differences. Additionally, the study did not investigate immunophenotype and functionality of TILs across molecular subtypes due to limited resources and research scope. Moreover, the study included only invasive BC before neoadjuvant therapy; however, those who received neoadjuvant therapy were not included. The absence of long-term follow-up data limits the assessment of the relationship between TILs and outcomes, such as recurrence and survival. Furthermore, the absence of a formal power calculation and relatively small sample size (n = 62) limit statistical power, particularly in early stages, non-ductal histologic types, and T stages. Consecutive sampling may have introduced selection bias, as stage IV patients who underwent mastectomy for diagnostic purposes could overrepresent surgically resected advanced disease. Finally, all participants were recruited exclusively from northwest Ethiopia; our findings may not be generalizable to other racial and diverse ethnic groups. Conclusion and recommendations This study found that low TIL scores were common in early stages, whereas intermediate and high TIL scores predominated in advanced stages. Specifically, median TIL percentages in stage IV and stage III were significantly higher than in stage II. TIL scores were positively correlated with TNM stage, nodal involvement, and histologic grade. These findings suggest that advanced tumors may harbor mutationally immunogenic features, and TILs in advanced stages may hypothetically reflect tumor-driven immune evasion rather than protective antitumor responses. Future studies should characterize immune cell subtypes and their spatial distribution within the TME. Furthermore, functional investigations are needed to elucidate mechanisms linking TILs to LN and distant metastasis. Further, future research assessing TIL distributions should include larger multi-center African studies and stratification by estrogen, progesterone, HER2-positive, and TNBC subtypes to address immune heterogeneity and ethnic variability in TILs.