We conducted a cross-sectional study in three hospitals located in two major cities in Cameroon, Yaoundé and Douala. In Yaoundé, data were collected from one tertiary referral hospital, Yaoundé General Hospital (YGH), and one private imaging centre, Centre Médical La Cathédrale (CMC). In Douala, data were obtained from one tertiary referral hospital, Laquintinie Hospital Douala (LHD).

The CT equipment available at the participating centers were as follows:

LHD: Siemens SOMATOM Perspective (128-slice);YGH: Siemens SOMATOM Go.Top (64-slice);CMC: Toshiba Aquilion (128-slice).

The study population comprised all patients who underwent thoraco-abdomino-pelvic (TAP) or abdominal CT examinations, irrespective of indication, between September 1 and December 31, 2024. Patients were eligible if they had undergone multiphasic TAP or abdominal CT with, at minimum, acquisition of both arterial and portal venous phases.

Exclusion criteria included major anatomical distortion (e.g., hepatic masses altering vascular anatomy, advanced cirrhosis with marked architectural distortion), inadequate vascular enhancement (severe motion artifacts affecting vascular visualization, poor contrast bolus timing with absence of clearly visible arterial or portal venous enhancement), technical failure of bolus tracking, incomplete coverage of the anatomical regions of interest, and repeat CT examinations in the same patient during the study period.

Data were collected consecutively and exhaustively. All examinations meeting the inclusion criteria were included in the analysis.

The data of interest included socio-demographic and radiologic variables. Socio-demographic characteristics comprised age and sex.

Radiologic evaluation focused on the hepatic arteries and portal vein with systematic identification and classification of anatomical variants.

For the hepatic arterial system, the origin of the common hepatic artery (CHA) was documented (celiac trunk, aorta, superior mesenteric artery [SMA], celiaco-mesenteric trunk, left gastric artery, or other). The number and origin of the right hepatic artery (RHA) and left hepatic artery (LHA) were recorded, including accessory or replaced branches (e.g., RHA arising from the SMA or LHA from the left gastric artery). Combined variants, the presence and origin of the middle hepatic artery, and other unclassified patterns (e.g., double hepatic artery, replaced hepatic artery proper from the SMA) were also assessed. Hepatic arterial anatomy was ultimately categorized according to the Michel’s classification that describes 10 patterns of hepatic arterial anatomy based on the origin of the right and left hepatic arteries. Type I corresponds to classic anatomy, with the common hepatic artery arising from the celiac trunk and dividing into right and left hepatic arteries. Variants include replaced or accessory left hepatic artery from the left gastric artery (types II and V), replaced or accessory right hepatic artery from the superior mesenteric artery (types III and VI), combined replaced or accessory patterns (types IV, VII, and VIII), and anomalous origin of the common hepatic artery from the superior mesenteric artery (type IX) or left gastric artery (type X). The classification distinguishes between “replaced” arteries, which substitute the normal branch, and “accessory” arteries, which coexist with it [15]. Any hepatic arterial pattern that did not strictly correspond to one of Michels’ classification types was categorized as “unclassified”.

For the portal venous system, the level of division (intrahepatic or extrahepatic) and the number of main branches were recorded. Specific branching patterns were documented according to the Covey classification [20], including the classical anatomy (type 1: main portal vein dividing into right and left portal veins, with right anterior and posterior branches arising from the right portal vein), right posterior portal vein as the first branch of the main portal vein (type 3), and separate segmental branches from the right portal vein to segment VII (type 4) or segment VI (type 5). Any portal vein pattern that did not strictly correspond to one of Covey classification types was categorized as “unclassified”.

A standardized multiphasic CT protocol was applied. Examinations were performed with the patient in the supine position, arms elevated, and the abdomen centered within the gantry. Tube voltage was ≤120 kVp, with tube current modulated using automatic exposure control.

Scanning extended from above the diaphragm to the pelvis, depending on the indication. The arterial phase typically covered the diaphragm to the iliac crest, while the portal venous phase extended from above the diaphragm to the symphysis pubis. Acquisition was performed in a craniocaudal direction.

Images were acquired with a field of view of approximately 350 mm (adjusted as needed), slice thickness ≤ 0.75 mm, and reconstruction interval ≤ 0.5 mm, using both soft tissue and bone algorithms.

Following an optional unenhanced acquisition, contrast-enhanced imaging was performed using 70 - 100 mL of iodinated contrast (approximately 0.1 mL/kg) administered at 3 - 5 mL/s, followed by a 30 - 40 mL saline flush. Bolus tracking was placed in the abdominal aorta. The arterial phase was acquired with minimal delay, and the portal venous phase at 60 - 80 seconds after contrast injection.

All acquisitions were performed during a single inspiratory breath-hold. Multiplanar reconstructions (axial, coronal, and sagittal) were systematically generated, with reconstructed slice thickness of 4 - 5 mm for soft tissue using 20% - 40% overlap.

CT examinations were reviewed on site using the available dedicated workstations and DICOM viewers at each center. Scans from Yaoundé General Hospital (YGH) were analyzed using the Siemens SOMATOM Workstation^® software and the Carestream DICOM^® viewer. Examinations from Centre Médical La Cathédrale (CMC) were reviewed with EvoluView^® and Weasis^® DICOM readers, while those from Laquintinie Hospital Douala (LHD) were assessed using the RadiAnt^® DICOM viewer.

All images were simultaneously reviewed and interpreted by a radiologist and a radiology assistant, with findings recorded by consensus.

Data were collected using Google Forms and exported to Microsoft Excel 2010 for initial data management. The dataset was subsequently transferred to IBM SPSS Statistics version 23 for statistical analysis.

Quantitative variables were summarized as means ± standard deviations (SD), and categorical variables as frequencies and proportions (%), presented in tables. Associations between categorical variables were assessed using the chi-square test. Statistical significance was defined as a two-sided p-value < 0.05.

Ethical approval was obtained from the Institutional Review Board of the Faculty of Medicine and Biomedical Sciences, University of Yaoundé I (Approval No. 0171/UY1/FMSB/VDRC/DAASR/CSD/emr).

The study population comprised 153 males and 286 females (male-to-female ratio, 1:1.8). Participants ranged in age from 4 to 95 years (mean 51.7 ± 16.3 years). Most patients were aged 41 - 60 years (205/439, 46.7%) (Figure 1).

Figure1. Age distribution of the study population.

Because of suboptimal vascular enhancement in some examinations, hepatic arterial anatomy was analysed in 420 patients and portal venous anatomy in 413 patients. Hepatic.

Among the 420 patients in whom hepatic arterial anatomy was assessable, standard anatomy was observed in 230 (54.8%, 95% CI: 50.0 - 59.5), whereas 190 (45.2%, 95% CI: 40.5 - 50.0) demonstrated variant patterns.

The prevalence of hepatic arterial variants was similar across the three study sites. Variant rates were 46.5% at YGH, 46.8% at CMC, and 42.9% at LDH. The inter-site comparison showed no statistically significant difference (p = 0.756), indicating that the observed variations are consistent across centres (Table 1).

Table1. Hepatic artery variant prevalence by site.

Inter-site chi square: X² = 0.559, P = 0.756.

Of the 413 patients evaluated for portal venous anatomy, 312 (75.5%, 95% CI: 71.2 - 79.4) had standard anatomy and 101 (24.5%, 95% CI: 20.6 - 28.8) exhibited anatomical variants.

The prevalence of portal venous variants showed moderate variation across sites, ranging from 20.9% at YGH to 29.8% at CMC, with LDH at 24.8%. Despite these differences the inter-site comparison did not reach statistical significance (p = 0.280) (Table 2).

Table2. Portal vein variant prevalence by site.

Inter-site chi square: X² = 2.549, P = 0.280.

Hepatic arterial anatomy was analyzed in 420 patients and classified according to the Michels classification (Table 3). Of the 10 variants described by Michel, nine (9) were identified in this study. The most frequent variants were type V (76/420, 18.1%), type II (46/420, 10.9%), and type III (22/420, 5.2%) (Figure 2, Figure 3).

Table3. Michel’s classification of hepatic arterial anatomy.

*LHA: left hepatic artery. LGA: left gastric artery. RHA: right hepatic artery. SMA: superior mesenteric artery. CHA: common hepatic artery.

Figure2. Volume rendering of a replaced LHA from the left gastric artery (arrow)—Michels’ Type V—Note presence of an accessory right renal artery (arrowhead).

Figure3. Volume rendering illustrating the CHA from the superior mesenteric artery—Type IX.

Figure4. Sagittal reconstruction image showing a celiaco-mesenteric trunk.

Eighteen patients exhibited hepatic arterial configurations that could not be assigned to any Michels type and were categorized as unclassified (Table 3). These included the presence of a celiaco-mesenteric trunk in 7 patients (Figure 4) and a double hepatic artery in 1 patient. Additional anomalous origins of the right hepatic artery were observed, arising from the celiac trunk in 11 patients (Figure 5), from a celiaco-mesenteric trunk in 1 patient, from the aorta in 1 patient, and from the common hepatic artery in 2 patients. Anomalous origins of the left hepatic artery from the celiac trunk (1 patient) and from the common hepatic artery (1 patient) were also identified. These variants were included in the overall prevalence of hepatic arterial variants but reported separately as they could not be mapped to a specific Michels type.

Figure5. Volume rendering of a RHA arising directly from the celiac trunk (also presence of a middle hepatic artery).

Of the 413 CT examinations analyzed for portal venous anatomy, 312 (75.5%) had the normal anatomy. According to Covey’s classification, the most frequent variant was a separate right posterior portal vein branch to segment VI (49/413, 11.9%) (Table 4).

Table4. Covey classification of portal vein anatomy.

*PV: portal vein. RPPV: right posterior portal vein. RPV: right portal vein.

Unclassified variants included quadrifurcation of the right portal vein, quadrifurcation of the main portal vein, and an atypical bifurcation in which the main portal vein divided into the right anterior and right posterior branches, with the left portal vein arising from the right anterior branch.

Among the 420 patients evaluated for hepatic arterial anatomy, standard anatomy was observed in 89 males (21.2%) and 141 females (33.6%), whereas variant anatomy was identified in 59 males (14.0%) and 131 females (31.2%). Although hepatic arterial variants were more frequently observed in females than in males, this difference was not statistically significant (p = 0.1).

Among the 413 patients assessed for portal venous anatomy, normal anatomy was identified in 106 males (25.7%) and 206 females (49.9%), whereas variant patterns were observed in 37 males (9.0%) and 64 females (15.5%). Although portal venous variants appeared more frequent in males, the difference was not statistically significant (p = 0.71).

There was also no observed correlation between portal venous and hepatic arterial variants (p = 0.277).

Regarding the hepatic arterial vasculature, normal anatomy was described as a CHA arising from the celiac trunk (which itself arose directly from the abdominal aorta) and giving off the proper hepatic artery and the gastroduodenal artery. The proper hepatic artery then had to bifurcate into the left and right hepatic arteries. Any variation of this pattern was termed an aberrant artery [15]. After analyzing 420 hepatic arterial systems, normal anatomy was seen in 54.8% of the population according to Michel’s classification, with variants accounting for 45.2%. This was similar to the original study carried out by Michel et al. in 1966 on cadavers [15], Muzenda in 2020 in South Africa [20] and Cawich et al. in 2021 in Trinidad and Tobago [17], who had a prevalence of 55%, 55% and 54.6% for normal anatomy respectively. However, this was lower than what was described in 2 systematic reviews by Löschner et al. in 2015 [21] and Noussios et al. in 2017 [22], with a prevalence of 72.2% and 81% respectively. This difference could be explained by the larger sample sizes in their studies (1297 and 19,103 patients respectively) compared to our sample of 420 patients analyzed for hepatic arterial anatomy. Also, their studies were systematic reviews assessing populations with different genetic backgrounds, eliminating any population-related bias.

Based on Michel’s classification, the most common hepatic arterial variant in our study was the accessory LHA from the left gastric artery (type V) with a prevalence of 18.1%. This variant was also the most common described by Muzenda [20], with a prevalence of 12.7%. However, this was in contrast to the studies carried out by Michel [15], Sambath [3], Fonseca-Neto [23] and Noussios [22], who had the type II variant (replaced LHA from left gastric artery) as the most common. This could be explained by the fact that Muzenda’s and our studies were based on sub-Saharan African populations (South Africa and Cameroon respectively), further supporting the theory that patterns of vascular variants are diverse probably due to genetic differences in different populations.

The types II (replaced LHA from left gastric artery) and III (replaced RHA from the superior mesenteric artery) variants were the next most common variants in our study, accounting for 11.0% and 5.2% respectively. The fact that the most common variants observed were both the accessory and replaced LHA from the left gastric artery (types V and II respectively accounting for 29.1% of all variants) signals the importance of verifying hepatic vasculature before gastric-related procedures in our context.

The type X variant (entire hepatic trunk arising from the left gastric artery) was not found in any of the patients. This is consistent with other studies like those of Sambath [3], Cawich [17] and Zaki [24]. Since it was described by Adachi in 1928 [25], only a handful of cadaveric and angiographic studies/case reports have described this variant in detail [26][27], revealing it to be a rare variant overall.

Portal venous anatomy was analyzed in 413 patients. Standard anatomy usually comprises bifurcation of the main portal vein (PV) at the hepatic hilum into LPV and RPV, with further bifurcation of the RPV into the RAPV and RPPV. This normal anatomy was observed in 75.5% of the study population, similar to that seen by Sureka et al. in 2015 [28] and Ulusoy et al. in 2020 [29]. They reported prevalences of 79.94% and 82.6% respectively. However, it was higher than in the studies by Covey [18] and Sambath [3], who showed a prevalence of 65% and 57.4% respectively. This difference could be explained by the relatively low sample sizes in the latter studies (200 and 350 patients respectively).

The most common portal vein variant in our study was the type V (separate branch of RPV to segment VI) seen in 11.9% of our study population, followed by trifurcation of the main PV in 7.7%. The type V variant was one of the least observed in the above studies, in which the type II (main PV trifurcation) was the most common variant followed by type III (RPPV as first branch of main PV).

No significant association was found between patient gender and the presence of portal venous variants (p = 0.71), consistent with previous reports by Ulusoy [29] and Sambath [30]. Similarly, hepatic arterial anatomy was not associated with gender (p = 0.1). This could be explained by the fact that there is no known embryological pathway involving both the hepatic vascular and the genital systems.

This study has some limitations. Its retrospective design may have introduced selection bias and limited control over imaging protocols across centers. A proportion of examinations were excluded because of suboptimal vascular enhancement, potentially affecting representativeness. Although multicentric, the study was hospital-based and may not fully reflect the general population. Finally, the use of CT rather than digital subtraction angiography, the reference standard, may have limited the detection of subtle vascular variants. Nevertheless, the study included a multicenter sample from two major urban settings and reflects real-world imaging practice. Moreover, multiphasic CT is the modality most used for preoperative planning in this context, making the findings directly applicable to surgical and interventional decision-making.