Trends in gastrointestinal cancers at a major tertiary hospital in Alexandria, Egypt: a 10-year retrospective study (2010–2019) Authors: Marium Mahmoud Abdulla, Osama Eisa Salama, Gamal El-Husseiny, Hany Hassan Ziady, Dalia El-Deeb Published: 2026 ABSTRACT BackgroundGastrointestinal (GI) cancers account for over one-quarter of global cancers, with marked regional variation in site distribution. In Egypt, contemporary data from large urban referral centres remain limited, particularly outside population-based cancer registries.MethodsA hospital-based retrospective review was conducted of patients diagnosed with GI cancers at Alexandria Main University Hospital between 2010 and 2019. Age, sex, residence, year of diagnosis, and cancer site were analysed using SPSS and Joinpoint regression to assess temporal trends in hospital-registered case counts. Sensitivity analyses examined yearly proportions of GI cancers relative to all cancers to account for fluctuations in hospital volume.ResultsAmong 21,949 cancers registered during the study period, 3,995 (18.2%) were GI cancers. Colorectal and liver cancers were the most frequently registered GI sites, followed by pancreatic cancer. Using raw counts, overall hospital-registered GI cancer trends showed no statistically significant change across the decade. Small intestinal cancers demonstrated a significant decline in hospital-registered cases during 2015–2019 (annual percent change = − 22.2%; 95% CI − 64.8 to − 5.8; P = 0.019). The mean age at diagnosis was 55.1 ± 12.2 years, and 27.4% of cases were early-onset (< 50 years). Joinpoint analysis by age-onset category showed a significant increase in late-onset cases, while early-onset cases declined after 2014. Males predominated overall (male-to-female ratio 1.38:1), particularly in liver cancers (2.71; 95% CI 2.40–3.10). Patients residing in Alexandria more frequently presented with colorectal and gastric cancers, whereas liver cancers were more common among patients residing outside Alexandria. Sex-specific Joinpoint analyses for major GI cancer sites showed no statistically significant temporal changes.ConclusionGI cancers represent a substantial proportion of hospital-registered malignancies at Alexandria Main University Hospital, with largely stable case-count trends over the past decade. Sensitivity analyses suggest that relative burden may change even when absolute counts appear stable. These findings reflect hospital presentation patterns rather than population incidence and highlight the need for expanded population-based cancer registration and regional surveillance in Egypt.Graphical Abstract Supplementary InformationThe online version contains supplementary material available at 10.1186/s12876-026-04803-8. Introduction Gastrointestinal (GI) cancers collectively account for nearly one quarter of the global cancer burden. In 2022, colorectal cancer (CRC) was the most common GI cancer site, and together with gastric cancer (GC), and liver cancers (LC) ranked among the world’s ten most common malignancies, while esophageal cancer (EC) and pancreatic cancer (PC) were also prominent contributors to cancer morbidity [1]. Similar patterns are observed in the Eastern Mediterranean Region, where CRC, LC, and GC also rank among the top causes of cancer incidence, albeit in a slightly different order than global figures [1]. In Egypt, estimates from GLOBOCAN 2022 identify LC as the most common malignancy, with CRC and PC also among the top ten sites [1]. Comparable patterns have been reported by the Egyptian National Population-Based Cancer Registry Program, which found LC and CRC to be among the leading cancer sites nationally [2]. However, both datasets rely on cancer registration from a limited number of governorates—Gharbiah for GLOBOCAN, and Damietta, Minya, and Aswan for the national registry. Consequently, these estimates may not fully represent the epidemiology of cancer in Egypt’s large metropolitan regions. Alexandria, the country’s second-largest city and a major referral hub for surrounding rural governorates, remains unrepresented in national cancer surveillance efforts. Given Egypt’s substantial burden of risk factors relevant to GI cancers, including historically high hepatitis C virus (HCV) prevalence, rising rates of non-alcoholic fatty liver disease, obesity, and increasingly westernized dietary patterns, understanding local GI cancer patterns is essential for service planning and early detection strategies [3–6]. Providing complementary evidence from a major urban center currently absent from national registry coverage is therefore of public health importance. Accordingly, this study aimed to describe the pattern and temporal trends of GI cancers registered at Alexandria Main University Hospital during 2010–2019. Methods Study setting and population This study adopted a hospital-based retrospective design. Medical records of registered GI cancers, available at the archive section of the Clinical Oncology and Nuclear Medicine Department in Alexandria Main University Hospital were reviewed to determine the pattern and temporal trends of GI cancers over the decade 2010–2019. A pre-determined data extraction sheet was used to collect information from medical records, including GI cancer site, age, sex, and residence. The International Classification of Diseases, 10th Revision (ICD-10) was used for cancer site coding in this study [7]. In the present study, colon, rectal, rectosigmoidal, and anal cancers were combined into CRC, in line with GLOBOCAN methodology [8]. The final cancer sites analyzed were EC, GC, small intestine cancer, CRC, liver and intrahepatic biliary tract cancers (hereafter referred to as LC), gallbladder cancer, other unspecified biliary tract cancers, and PC. Age at cancer diagnosis was categorized into early-onset (< 50 years), average-onset (50–<65 years), and late-onset (≥ 65 years) [9, 10]. As Alexandria is predominantly urban, while the other governorates served by the hospital are largely rural, residence was classified as Alexandria (urban) versus outside Alexandria (rural). This binary classification is an operational definition and may misclassify some peri-urban areas or migratory populations. Cases were included if they had a confirmed histopathological diagnosis or had initiated cancer-directed treatment. For LC, cases without histopathology were included if clinical and imaging criteria strongly suggested hepatocellular carcinoma (positive HCV serology, elevated alpha-fetoprotein (AFP), and LIRADS-5 imaging classification). No exposures, predictors, or effect modifiers were defined, as the study aimed to describe patterns of hospital-registered cases. Additionally, patients’ names were initially collected in Arabic, exactly as recorded in the medical files, solely to facilitate duplicate identification and were subsequently removed to maintain confidentiality. The study included all available GI cancer cases registered at the department from 2010 to 2019 that met the inclusion criteria. No formal sample size calculation was performed, as the study aimed to analyze the complete hospital-based dataset over the decade. All records registered during 2010–2019 (n = 21,949) were reviewed to identify GI cancer cases. Data related to GI cancers were extracted from physical medical records stored in the department archive during the period from January to October 2024. Records were logged into Excel, facilitating the identification of duplicates by name. Potential duplicates were verified using national ID number, phone number, or address, and any confirmed duplicates were excluded. A total of 4,323 medical records were initially identified. However, 328 records were excluded due to inconclusive diagnosis (n = 261, 79.6%), duplication (n = 64, 19.5%), or biopsy revealing carcinoma in situ (n = 3, 0.9%), which is coded separately as D00–D01 in ICD-10 [7]. Thus, the final analytical sample comprised 3,995 GI cancer records. This study was designed and reported in accordance with the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) guidelines to ensure transparency and completeness in reporting observational research. A filled-out STROBE Statement—checklist is as provided as Supplementary file 1. Bias and data quality To reduce selection bias, only histopathology-confirmed cases or patients who received cancer-directed treatment were included; for LC, cases meeting clinical and imaging criteria (HCV-positive, elevated AFP, LIRADS-5) were also accepted. Data were extracted using a standardized sheet by a single data collector to minimize measurement error. Patient names were entered in Arabic exactly as recorded in the medical files, and potential duplicates were rigorously verified using national ID, phone number, or address, with any confirmed duplicates excluded. Residence classification was applied consistently, and cases with missing residence information (n = 86) were included in overall analyses but excluded from residence-stratified analyses. Confidence intervals were calculated to reflect statistical uncertainty, particularly for cancer sites with small annual counts. Ethical committee approval Ethics approval was obtained from the Institutional Review Board (IRB No: 00012098) at the Faculty of Medicine, Alexandria University. As part of a thesis, consent waiver was granted for this part of the study, by the ethical committee due to the retrospective nature of the study. Statistical methods Data were analyzed using SPSS version 23.0 and the Joinpoint Regression Program version 5.4.0 (National Cancer Institute, USA) [11]. Continuous variables were summarized as mean ± standard deviation (SD) and compared using Kruskal–Wallis tests, after the Shapiro–Wilk test confirmed non-normality. Age at cancer diagnosis was analyzed both as a continuous variable and as a categorical variable, (grouped into early, average, and late-onset). Year of registration was treated as a continuous variable for trend analyses in Joinpoint regression. Categorical variables were summarized as frequencies and percentages, and group comparisons were conducted using Chi-square tests. Effect sizes were calculated using ordinal eta-squared (η²) for quantitative variables and Cramer’s V for categorical variables. A significance level of P ≤ 0.05 was applied to all analyses, and 95% confidence intervals (CI) were calculated. Joinpoint settings Temporal trends were examined overall and by subgroups (age-onset category, sex, and residence) using Joinpoint regression to estimate annual percent change (APC) and average annual percent change (AAPC). Annual case counts by age-onset category, sex, and residence are provided in Supplementary file 2 to allow assessment of temporal stability underlying Joinpoint analyses. The maximum number of joinpoints was restricted to one, as recommended given the ten annual data points [12]. The minimum number of observations per segment (default = two) was retained, and log-transformation was applied as estimation of relative change and APCs was of interest. The Weighted Bayesian Information Criterion was used for model selection to allow flexibility in identifying the optimal model [12]. For cancer sites with sparse annual counts, estimates were interpreted cautiously due to potential instability. For residence-stratified analyses, cases with missing residence information (n = 86) were excluded. Sensitivity analyses were conducted by calculating annual proportions of GI cancers relative to all cancers (total GI / total cancers) and site-specific proportions (GI site / total cancers) to account for potential fluctuations in annual hospital case volume. Standard errors were calculated for each proportion. Joinpoint models assumed Poisson variance for count-based analyses, while calculated standard errors were used for proportion-based analyses to account for heteroscedasticity. These analyses examined temporal trends in the relative representation of GI cancers. Detailed 95% CIs for APC and AAPC are provided in Supplementary files 3, 5, and 6, while Supplementary file 4 presents APC estimates from sensitivity analyses using proportions. Handling of confounding and missing data As this was a descriptive, hospital-based study without regression modelling, no formal adjustment for confounding was performed. Analyses were primarily stratified by key variables (age, sex, residence, and cancer site) to reduce potential confounding by these factors. Missing data were negligible for most variables, except residence (n = 86; 2.2%), which was handled by complete-case analysis in residence-stratified models. No imputation was performed due to the retrospective nature of the data and lack of auxiliary variables. The potential impact of missing data on estimates is addressed in the limitations section. Results Temporal trends and annual percent changes in gastrointestinal cancers Overall trend A total of 3,995 GI cancer records were identified during the decade 2010–2019 and included in the analysis. The process of case selection is shown in Fig. 1. Overall, GI cancers accounted for 18.2% of all cancer sites. As shown in Fig. 2, GI cancer cases were highest in 2016 (n = 472) and lowest in 2019 (n = 360). Joinpoint regression showed a non-significant increase in APC (+ 2.74%) during 2010–2016, followed by a non-significant decrease (–5.57%) during 2016–2019. The AAPC over the entire decade was not statistically significant (–0.11%). Details of 95% CIs and P-values are provided in Supplementary file 3. Fig. 1STROBE participant flow chart STROBE participant flow chart Fig. 2Joinpoint model for total and gastrointestinal cancers registered at Alexandria Main University Hospital in the decade 2010–2019 Joinpoint model for total and gastrointestinal cancers registered at Alexandria Main University Hospital in the decade 2010–2019 Trends by cancer site As shown in Fig. 3, an initial increase followed by a decrease was observed in liver and intrahepatic biliary tract cancers; however, these changes were not statistically significant. The only significant APC was observed in small intestine cancers, which showed a significant decline during 2015–2019 (APC = − 22.21%, 95% CI: − 64.83 to − 5.77, P = 0.019), despite small annual case counts (< 30 in several years). No cancer site demonstrated a statistically significant AAPC. Detailed APCs, AAPCs, 95% CIs, and P-values are provided in Supplementary file 3. Fig. 3Joinpoint model for high-incidence (a) and low-incidence gastrointestinal cancers (b), registered at Alexandria Main University Hospital in the decade 2010–2019 Joinpoint model for high-incidence (a) and low-incidence gastrointestinal cancers (b), registered at Alexandria Main University Hospital in the decade 2010–2019 Sensitivity analysis using proportions In sensitivity analyses accounting for annual hospital caseload fluctuations, Joinpoint regression was applied to yearly proportions of GI cancers relative to all registered cancers. This analysis revealed a significant biphasic trend in the proportional burden of GI cancers, with a significant increase from 2010 to 2012 (APC = + 15.84%, 95% CI: 0.21 to 34.44, P = 0.045), followed by a significant decline from 2012 to 2019 (APC = − 3.50%, 95% CI: − 15.41 to − 0.74, P = 0.036). At the site-specific level, small intestine cancers were the only GI site showing a statistically significant proportional trend, with a significant decline during 2015–2019. All other GI cancer sites showed no significant proportional trends. Detailed APC estimates and 95% CIs are provided in Supplementary file 4. Demographics by year of diagnosis Demographic characteristics by year of diagnosis are summarized in Table 1. The mean age at diagnosis was 55.1 ± 12.2 years, ranging from 54.0 years in 2011 to 56.4 years in 2018. Although differences in mean age across years were statistically significant (Kruskal–Wallis P = 0.046), the effect size was negligible (η² = 0.002). Table 1Demographic characteristics of patients with gastrointestinal cancers registered at Alexandria Main University Hospital, 2010–2019, by year2010201120122013201420152016201720182019TotalP-value (effect size)#Total n (%)371 (9.29)422 (10.56)382 (9.56)366 (9.16)401 (10.04)401 (10.04)472 (11.81)421 (10.54)399 (9.99)360 (9.01)3,995 (100.00)Age (mean ± SD)54.75 ± 12.7654.02 ± 12.0654.56 ± 12.1154.53 ± 12.3954.28 ± 12.7755.57 ± 11.5855.32 ± 11.8156.09 ± 11.8656.42 ± 12.1155.81 ± 12.8955.13 ± 12.230.046* (0.002)[95% CI for mean]53.44–56.0652.86–55.1753.34–55.7853.26–55.8153.03–55.5454.43–56.7254.25–56.3954.95–57.2455.20–57.6454.46–57.1654.75–55.51Age-onset category No. (%)0.004* (0.004)Early (< 50)102 (27.49)123 (29.15)111 (29.06)113 (30.87)122 (30.42)116 (28.93)119 (25.21)98 (23.28)94 (23.56)97 (26.94)1095 (27.41)Average (50-<65)185 (49.87)226 (53.55)191 (50.00)174 (47.54)197 (49.13)195 (48.63)257 (54.45)211 (50.12)184 (46.12)170 (47.22)1990 (49.81)Late (≥ 65)84 (22.64)73 (17.30)80 (20.94)79 (21.58)82 (20.45)90 (22.44)96 (20.34)112 (26.60)121 (30.33)93 (25.83)910 (22.78)Sex No. (%)0.042* (0.042)Male226 (60.90)257 (60.90)224 (58.60)217 (59.30)234 (58.40)224 (55.90)243 (51.50)243 (57.70)250 (62.70)196 (54.40)2314 (57.92)Female145 (39.10)165 (39.10)158 (41.40)149 (40.70)167 (41.60)177 (44.10)229 (48.50)178 (42.30)149 (37.30)164 (45.60)1681 (42.08)Male to female ratio1.551.551.421.461.401.261.061.361.671.191.38[95% CI for Male to female ratio]1.27–1.921.28–1.891.16–1.741.18–1.791.15–1.711.04–1.540.89–1.271.13–1.661.37–2.050.97–1.471.29–1.47Residence No. (%)$0.294 (≈ 0.000)Alexandria (urban)224 (60.40)250 (59.20)223 (58.40)219 (59.80)216 (53.90)238 (59.40)265 (56.10)235 (55.80)225 (56.40)221 (61.40)2316 (57.97)Outside (rural)139 (37.50)170 (40.30)157 (41.10)144 (39.30)181 (45.10)157 (39.20)199 (42.20)177 (42.00)146 (46.60)123 (34.20)1593 (39.87)Urban to rural ratio1.611.471.421.521.191.521.331.331.541.801.45[95% CI for urban to rural ratio]1.30–1.991.21–1.781.16–1.741.23–1.880.98–1.451.24–1.851.11–1.601.09–1.611.25–1.901.44–2.231.36–1.55P-values compare across years using Kruskal-Wallis (for age) and χ² (for categorical variables)*: significant P-value (< 0.05)#: Effect size calculated using Cramer’s V for categorical data, and ordinal Eta-squared (η²) for continuous data$: Residence is classified as Alexandria (urban) vs outside Alexandria (rural). This is an operational definition and may misclassify some peri-urban or migratory populations. Missing residence data: 86 cases (2.2% of total), thus total urban and rural does not sum to 100% Table 1Demographic characteristics of patients with gastrointestinal cancers registered at Alexandria Main University Hospital, 2010–2019, by year2010201120122013201420152016201720182019TotalP-value (effect size)#Total n (%)371 (9.29)422 (10.56)382 (9.56)366 (9.16)401 (10.04)401 (10.04)472 (11.81)421 (10.54)399 (9.99)360 (9.01)3,995 (100.00)Age (mean ± SD)54.75 ± 12.7654.02 ± 12.0654.56 ± 12.1154.53 ± 12.3954.28 ± 12.7755.57 ± 11.5855.32 ± 11.8156.09 ± 11.8656.42 ± 12.1155.81 ± 12.8955.13 ± 12.230.046* (0.002)[95% CI for mean]53.44–56.0652.86–55.1753.34–55.7853.26–55.8153.03–55.5454.43–56.7254.25–56.3954.95–57.2455.20–57.6454.46–57.1654.75–55.51Age-onset category No. (%)0.004* (0.004)Early (< 50)102 (27.49)123 (29.15)111 (29.06)113 (30.87)122 (30.42)116 (28.93)119 (25.21)98 (23.28)94 (23.56)97 (26.94)1095 (27.41)Average (50-<65)185 (49.87)226 (53.55)191 (50.00)174 (47.54)197 (49.13)195 (48.63)257 (54.45)211 (50.12)184 (46.12)170 (47.22)1990 (49.81)Late (≥ 65)84 (22.64)73 (17.30)80 (20.94)79 (21.58)82 (20.45)90 (22.44)96 (20.34)112 (26.60)121 (30.33)93 (25.83)910 (22.78)Sex No. (%)0.042* (0.042)Male226 (60.90)257 (60.90)224 (58.60)217 (59.30)234 (58.40)224 (55.90)243 (51.50)243 (57.70)250 (62.70)196 (54.40)2314 (57.92)Female145 (39.10)165 (39.10)158 (41.40)149 (40.70)167 (41.60)177 (44.10)229 (48.50)178 (42.30)149 (37.30)164 (45.60)1681 (42.08)Male to female ratio1.551.551.421.461.401.261.061.361.671.191.38[95% CI for Male to female ratio]1.27–1.921.28–1.891.16–1.741.18–1.791.15–1.711.04–1.540.89–1.271.13–1.661.37–2.050.97–1.471.29–1.47Residence No. (%)$0.294 (≈ 0.000)Alexandria (urban)224 (60.40)250 (59.20)223 (58.40)219 (59.80)216 (53.90)238 (59.40)265 (56.10)235 (55.80)225 (56.40)221 (61.40)2316 (57.97)Outside (rural)139 (37.50)170 (40.30)157 (41.10)144 (39.30)181 (45.10)157 (39.20)199 (42.20)177 (42.00)146 (46.60)123 (34.20)1593 (39.87)Urban to rural ratio1.611.471.421.521.191.521.331.331.541.801.45[95% CI for urban to rural ratio]1.30–1.991.21–1.781.16–1.741.23–1.880.98–1.451.24–1.851.11–1.601.09–1.611.25–1.901.44–2.231.36–1.55P-values compare across years using Kruskal-Wallis (for age) and χ² (for categorical variables)*: significant P-value (< 0.05)#: Effect size calculated using Cramer’s V for categorical data, and ordinal Eta-squared (η²) for continuous data$: Residence is classified as Alexandria (urban) vs outside Alexandria (rural). This is an operational definition and may misclassify some peri-urban or migratory populations. Missing residence data: 86 cases (2.2% of total), thus total urban and rural does not sum to 100% Demographic characteristics of patients with gastrointestinal cancers registered at Alexandria Main University Hospital, 2010–2019, by year P-values compare across years using Kruskal-Wallis (for age) and χ² (for categorical variables) *: significant P-value (< 0.05) #: Effect size calculated using Cramer’s V for categorical data, and ordinal Eta-squared (η²) for continuous data $: Residence is classified as Alexandria (urban) vs outside Alexandria (rural). This is an operational definition and may misclassify some peri-urban or migratory populations. Missing residence data: 86 cases (2.2% of total), thus total urban and rural does not sum to 100% The distribution of age-onset categories also differed significantly across years (χ² P = 0.004); however, the effect size was extremely small (Cramer’s V = 0.004), indicating minimal practical variation. Sex distribution varied significantly across years (χ² P = 0.042), with a weak association (Cramer’s V = 0.042). Males predominated throughout the study period, with male-to-female ratios ranging from 1.06 in 2016 to 1.67 in 2018. Residence did not vary significantly across years (χ² P = 0.294), and the effect size was negligible (Cramer’s V < 0.001). The urban-to-rural ratio ranged from 1.19 in 2014 to 1.80 in 2019. Demographics by cancer site As shown in Table 2, mean age at diagnosis differed significantly by cancer site (P < 0.001), with the highest mean age observed in liver and intrahepatic biliary tract cancers (58.9 years) and gallbladder cancer (58.2 years), and the lowest in small intestine cancers (50.6 years). Despite statistical significance, the effect size was small (η² = 0.049). Table 2Demographic characteristics of patients with gastrointestinal cancers registered at Alexandria Main University Hospital, 2010–2019, by cancer siteEsophagus (C15)Stomach(C16)Small intestine (C17)Colorectal (C18-C21)Liver & intra-hepatic biliary tract (C22)Gallbladder (C23)Other unspecified biliary tract (C24)Pancreas (C25)P-value (effect size)Total n (%)150 (3.75)514 (12.87)108 (2.70)1334 (33.39)1193 (29.86)56 (1.40)102 (2.55)538 (13.47)Age (mean ± SD)55.64 ± 13.6153.33 ± 12.8650.56 ± 13.7652.10 ± 13.7658.88 ± 9.0058.20 ± 10.5155.07 ± 9.7056.64 ± 10.95< 0.001* (0.049)[95% CI for mean]53.43–57.8552.21–54.4547.93–53.1851.36–52.8458.37–59.4055.38–61.0153.15–56.9855.71–57.57Age-onset category No. (%)< 0.001* (≈ 0.000)Early (< 50)43 (28.57)192 (37.35)48 (44.44)524 (39.28)134 (11.23)9 (16.07)23 (22.55)122 (22.68)Average (50-<65)66 (44.00)215 (41.83)40 (37.04)550 (41.23)733 (61.44)32 (57.14)63 (61.76)291 (54.09)Late (≥ 65)41 (27.33)107 (20.82)20 (18.52)260 (19.49)326 (27.33)15 (26.79)16 (15.69)125 (23.23)Sex No. (%)< 0.001* (≈ 0.000)Male88 (3.80)262 (11.30)51 (2.20)623 (26.90)872 (37.70)18 (0.80)59 (2.50)341 (14.70)Female62 (3.70)252 (15.00)57 (3.40)711 (42.30)321 (19.10)38 (2.30)43 (2.60)197 (11.70)Male to female ratio1.421.040.890.872.710.471.371.73[95% CI for Male to female ratio]1.02–1.960.87–1.240.61–1.300.79–0.982.40–3.080.27–0.830.93–2.031.45–2.06Residence No. (%)$< 0.001* (≈ 0.000)Alexandria (urban)65 (2.80)366 (15.80)71 (3.10)894 (38.60)529 (22.80)38 (1.60)57 (2.50)296 (12.80)Outside (rural)77 (4.80)138 (8.70)37 (2.30)411 (25.80)632 (39.70)18 (1.10)43 (2.70)237 (14.90)Urban to rural ratio0.842.651.922.180.842.111.331.25[95% CI for urban to rural ratio]0.60–1.172.18–3.221.28–2.851.94–2.440.75–0.941.20–3.690.89–1.0961.05–1.48Colorectal (C18-C21) includes colon, rectosigmoid, rectal and anal cancersP-values compare across cancer sites using Kruskal-Wallis (for age) and χ² (for categorical variables)*: significant P-value (< 0.05)#: Effect size calculated using Cramer’s V for categorical data, and ordinal Eta-squared (η²) for continuous data$: Residence is classified as Alexandria (urban) vs. outside Alexandria (rural). This is an operational definition and may misclassify some peri-urban or migratory populations. Missing residence data: 86 cases (2.2% of total), thus total urban and rural does not sum to 100% Table 2Demographic characteristics of patients with gastrointestinal cancers registered at Alexandria Main University Hospital, 2010–2019, by cancer siteEsophagus (C15)Stomach(C16)Small intestine (C17)Colorectal (C18-C21)Liver & intra-hepatic biliary tract (C22)Gallbladder (C23)Other unspecified biliary tract (C24)Pancreas (C25)P-value (effect size)Total n (%)150 (3.75)514 (12.87)108 (2.70)1334 (33.39)1193 (29.86)56 (1.40)102 (2.55)538 (13.47)Age (mean ± SD)55.64 ± 13.6153.33 ± 12.8650.56 ± 13.7652.10 ± 13.7658.88 ± 9.0058.20 ± 10.5155.07 ± 9.7056.64 ± 10.95< 0.001* (0.049)[95% CI for mean]53.43–57.8552.21–54.4547.93–53.1851.36–52.8458.37–59.4055.38–61.0153.15–56.9855.71–57.57Age-onset category No. (%)< 0.001* (≈ 0.000)Early (< 50)43 (28.57)192 (37.35)48 (44.44)524 (39.28)134 (11.23)9 (16.07)23 (22.55)122 (22.68)Average (50-<65)66 (44.00)215 (41.83)40 (37.04)550 (41.23)733 (61.44)32 (57.14)63 (61.76)291 (54.09)Late (≥ 65)41 (27.33)107 (20.82)20 (18.52)260 (19.49)326 (27.33)15 (26.79)16 (15.69)125 (23.23)Sex No. (%)< 0.001* (≈ 0.000)Male88 (3.80)262 (11.30)51 (2.20)623 (26.90)872 (37.70)18 (0.80)59 (2.50)341 (14.70)Female62 (3.70)252 (15.00)57 (3.40)711 (42.30)321 (19.10)38 (2.30)43 (2.60)197 (11.70)Male to female ratio1.421.040.890.872.710.471.371.73[95% CI for Male to female ratio]1.02–1.960.87–1.240.61–1.300.79–0.982.40–3.080.27–0.830.93–2.031.45–2.06Residence No. (%)$< 0.001* (≈ 0.000)Alexandria (urban)65 (2.80)366 (15.80)71 (3.10)894 (38.60)529 (22.80)38 (1.60)57 (2.50)296 (12.80)Outside (rural)77 (4.80)138 (8.70)37 (2.30)411 (25.80)632 (39.70)18 (1.10)43 (2.70)237 (14.90)Urban to rural ratio0.842.651.922.180.842.111.331.25[95% CI for urban to rural ratio]0.60–1.172.18–3.221.28–2.851.94–2.440.75–0.941.20–3.690.89–1.0961.05–1.48Colorectal (C18-C21) includes colon, rectosigmoid, rectal and anal cancersP-values compare across cancer sites using Kruskal-Wallis (for age) and χ² (for categorical variables)*: significant P-value (< 0.05)#: Effect size calculated using Cramer’s V for categorical data, and ordinal Eta-squared (η²) for continuous data$: Residence is classified as Alexandria (urban) vs. outside Alexandria (rural). This is an operational definition and may misclassify some peri-urban or migratory populations. Missing residence data: 86 cases (2.2% of total), thus total urban and rural does not sum to 100% Demographic characteristics of patients with gastrointestinal cancers registered at Alexandria Main University Hospital, 2010–2019, by cancer site Colorectal (C18-C21) includes colon, rectosigmoid, rectal and anal cancers P-values compare across cancer sites using Kruskal-Wallis (for age) and χ² (for categorical variables) *: significant P-value (< 0.05) #: Effect size calculated using Cramer’s V for categorical data, and ordinal Eta-squared (η²) for continuous data $: Residence is classified as Alexandria (urban) vs. outside Alexandria (rural). This is an operational definition and may misclassify some peri-urban or migratory populations. Missing residence data: 86 cases (2.2% of total), thus total urban and rural does not sum to 100% Age-onset category varied significantly across sites (P < 0.001), but the effect size was very weak (Cramer’s V ≈ 0.000). Small intestine cancers and CRC had the highest proportions of early-onset disease, whereas esophageal and LC had the highest proportions of late-onset disease. Sex distribution differed significantly by site (P < 0.001), though with a very small effect size (Cramer’s V ≈ 0.000). Most GI cancers showed male predominance, except small intestine, colorectal, and gallbladder cancers, where females predominated. Liver and intrahepatic biliary tract cancers showed the strongest male predominance (male-to-female ratio = 2.71; 95% CI: 2.40–3.08). Residence was significantly associated with cancer site (P < 0.001), though the association was weak (Cramer’s V ≈ 0.000). Urban residents more frequently presented with gastric and colorectal cancers, whereas LC was more common among patients residing outside Alexandria (urban-to-rural ratio = 0.84; 95% CI: 0.75–0.94). Annual percent changes by age, sex and residence As shown in Fig. 4, GI cancer APC decreased significantly in the early-onset group during 2014–2019 (APC = − 5.14%), while a significant increase was observed in the late-onset group (AAPC = 4.43%). Fig. 4Joinpoint model for gastrointestinal cancers registered at Alexandria Main University Hospital in the decade 2010–2019 by age of onset Joinpoint model for gastrointestinal cancers registered at Alexandria Main University Hospital in the decade 2010–2019 by age of onset As shown in Fig. 5, APC among males was − 2.13%, which was not statistically significant. Among females, APC was + 5.78% during 2010–2016 and − 8.63% during 2016–2019, with no significant AAPC for either sex. Fig. 5Joinpoint model for gastrointestinal cancers registered at Alexandria Main University Hospital in the decade 2010–2019 by sex Joinpoint model for gastrointestinal cancers registered at Alexandria Main University Hospital in the decade 2010–2019 by sex As shown in Fig. 6, APC by residence was significant only among patients residing outside Alexandria during 2017–2019 (APC = − 18.98%, 95% CI: − 36.51 to − 1.83, P = 0.036). AAPC by residence was not statistically significant. Detailed estimates are provided in Supplementary file 5. Fig. 6Joinpoint model for gastrointestinal cancers registered at Alexandria Main University Hospital in the decade 2010–2019 by residence Joinpoint model for gastrointestinal cancers registered at Alexandria Main University Hospital in the decade 2010–2019 by residence Annual percent changes in most common cancers by sex Joinpoint analysis of the most common GI cancer sites revealed no statistically significant APCs for LC among males, nor for CRC among males or females. Detailed APC estimates and 95% CIs are provided in Supplementary file 6. Discussion The present retrospective analysis showed that GI cancers accounted for 18.2% of all malignancies registered at Alexandria Main University Hospital during 2010–2019, with CRC and LC being the most frequent sites. These findings can be contextualized against earlier data from the same institution (1987–1996), when GI cancers constituted only 8.4% of all hospital-registered malignancies, indicating a substantial increase in the institutional case burden over the intervening decades [13]. Joinpoint regression revealed a stable overall, site- and sex-specific trends across the decade, with no statistically significant APC or AAPC. These findings reflect stability in hospital case counts rather than population-level incidence or true epidemiological plateaus. Such stability may also reflect the absence of organised cancer screening, and constrained diagnostic capacity, both of which can contribute to relatively unchanged hospital case loads despite evolving risk factors. Compared with the present study, earlier data in the same setting, reported rising counts of CRC, LC and PC, with a declining count in GC and EC [13], although formal trend modelling was not performed in that period, limiting direct comparison. While global studies report divergent GI cancer incidence trends [14, 15], most describe APC values without formal hypothesis testing. In contrast, the present study applied Joinpoint regression with confidence intervals, allowing formal statistical testing of temporal changes in hospital-registered cases, and confirming that observed fluctuations did not reach statistical significance. This analytic approach strengthens inference regarding temporal stability within this hospital setting but should not be interpreted as evidence of stable population incidence. The sensitivity analysis using proportions of GI cancers relative to total cancers partially addresses annual hospital-volume fluctuations. While total GI cancer counts showed no significant APC, proportional analysis demonstrated a significant increase in early years followed by a significant decline thereafter, indicating changes in relative hospital burden rather than absolute case volume. Small intestine cancers remained significant in both count- and proportion-based analyses, supporting the robustness of this finding, although the sparse annual counts for this site may still limit precision and widen confidence intervals. Overall, the temporal patterns identified in this study represent trends in hospital-registered case counts and proportions rather than population-based incidence. Accordingly, observed APC and AAPC estimates should be interpreted as reflecting patterns of case presentation to Alexandria Main University Hospital, influenced by referral pathways, diagnostic practices, and service availability. Any proposed etiologic explanations remain speculative and require confirmation using population-based registries with appropriate denominators and individual-level exposure data. CRC was the most frequently registered GI cancer, consistent with findings from the same centre during 1987–1996 [13], , followed by LC, PC, GC, and EC. Although hospital-based, this distribution broadly aligns with national estimates from GLOBOCAN 2022 and the National Population-Based Cancer Registry Program [1, 2], while highlighting Alexandria as a major urban referral hub that remains underrepresented in national surveillance efforts. Differences from national patterns may reflect regional exposures, referral dynamics, or case-mix, rather than true incidence variation. The mean age at diagnosis of GI cancers in this cohort was 55.1 years, considerably higher than 48.5 years reported from the same setting during 1987–1996 [13]. This shift likely reflects demographic ageing. Joinpoint analysis showed a significant increase in late-onset cases, consistent with ageing of the patient population. The observed decline in early-onset cases after 2014 should be interpreted cautiously, as it may reflect changes in referral patterns or healthcare utilisation rather than a true reduction in disease occurrence. Any attribution to HCV treatment effects remains hypothetical, as individual-level infection or treatment data were not available. A male predominance was observed across most GI cancers in this cohort, particularly for LC and PC. This finding is consistent with global evidence, and explained on biological and behavioral grounds [16–18]. In contrast, gallbladder cancer showed a female predominance, consistent with gallstone-related risk [19, 20]. The female predominance observed for CRC contrasts with earlier institutional and international data [13, 21–23] and may reflect hospital-specific referral patterns or differential obesity burden rather than a true reversal of sex-specific incidence. CRC and GC were more frequent among patients residing in Alexandria, while LC was more frequent among those residing outside Alexandria. Residence was operationally defined as Alexandria versus outside Alexandria, which may misclassify peri-urban or migratory populations and oversimplify exposure patterns. Attributing these differences to urban lifestyle factors (e.g., obesity, westernised diet) [5, 6, 24] or rural exposures (e.g., HCV prevalence, agricultural chemicals) [18, 25], remains speculative. The late-period decline in rural cases likely reflects changes in referral pathways or disease management rather than confirmed population-level risk reduction. Strengths of the present study include a decade-long dataset from a major tertiary centre, a large and diverse cohort, and use of Joinpoint regression with confidence intervals to formally test hospital case trends. Limitations of the study This study has several limitations that should be considered when interpreting the findings. First, as a hospital-based retrospective analysis, selection bias is possible. Only cases registered at Alexandria Main University Hospital were included, and inclusion required histopathological confirmation or initiation of cancer-directed treatment; for LC, cases meeting imaging and clinical criteria (HCV-positive, elevated AFP, LIRADS-5) were also accepted. These inclusion criteria, together with potential referral and case-mix effects, may not reflect population-level incidence. Second, measurement bias may have occurred due to the retrospective extraction of data from paper medical records by a single data collector, with potential incompleteness of documentation and variability in the consistency and quality of records over the studied decade. The duplicate identification and removal process raises the possibility of both missed and erroneously excluded cases. Patient residence recording using a binary classification (Alexandria vs. outside Alexandria) may misclassify peri-urban or migratory populations. Missing residence information (n = 86; 2.2%) was excluded from residence-stratified analyses but included in overall analyses, which may introduce bias if the missingness is not random and could affect the validity of residence-based comparisons. Other critical variables—including tumor stage, viral hepatitis status, H. pylori infection, obesity, diabetes, smoking, and alcohol consumption—were unavailable, representing unmeasured confounders that limit interpretation of trends and prevent assessment of etiologic factors. Third, statistical precision was limited for some cancer sites due to small annual counts, resulting in wide confidence intervals for Joinpoint estimates. The large number of subgroup analyses conducted without adjustment for multiple testing may increase the likelihood of false-positive findings. Finally, the study’s generalizability is limited. Findings reflect patterns of hospital-registered cases and are influenced by referral pathways, diagnostic capacity, and service availability, rather than true population-based incidence. Consequently, observed trends should be interpreted cautiously, and extrapolation to the broader population is not warranted. Despite these limitations, the study highlights key temporal trends and patterns of GI cancers at a major tertiary center over a decade. The observed increases or decreases in specific cancer types should be interpreted cautiously, considering potential biases, missing clinical variables, and the hospital-based nature of the dataset. Comparisons with previous national and regional studies provide contextual support but cannot fully overcome the limitations inherent to retrospective hospital data. Conclusion GI cancers constituted a substantial proportion of malignancies managed at Alexandria Main University Hospital during 2010–2019, with CRC and LC being the most common types. Joinpoint regression demonstrated no statistically significant temporal changes in overall or site-specific hospital case trends, indicating relative stability in hospital-registered GI cancer burden over the past decade. While the observed distribution partially overlaps with national estimates for Egypt, the findings should be interpreted as reflecting referral-based patterns rather than population incidence, and they also highlight regional characteristics, including a higher representation of CRC, younger age at presentation, and distinct sex- and residence-based patterns. These results underscore the importance of expanding population-based cancer registration to include Alexandria and other large metropolitan centers, incorporating staging and outcomes, which is essential for accurately characterizing national cancer trends, and informing evidence-based cancer control strategies, screening policies, and resource allocation. Supplementary Information Supplementary Material 1. Supplementary Material 1. Supplementary Material 2. Supplementary Material 2. Supplementary Material 3. Supplementary Material 3. Supplementary Material 4. Supplementary Material 4. Supplementary Material 5. Supplementary Material 5. Supplementary Material 6. Supplementary Material 6.