Prognostic factors and survival outcomes in colorectal cancer: insights from a Ghanaian cohort

Abstract

Background

Colorectal cancer (CRC) incidence in Ghana has increased significantly in recent years. Despite medical advances, survival rates remain low, with challenges in CRC screening implementation contributing to late-stage diagnoses and poor outcomes.

Methods

This study updates CRC survival estimates and identifies key prognostic factors influencing outcomes in a Ghanaian cohort. We conducted a retrospective analysis of 281 CRC patients diagnosed between 2016 and 2022 at Komfo Anokye Teaching Hospital in Kumasi, Ghana. Data on patient demographics, tumour characteristics, and clinical variables were collected. Survival estimates were calculated using the Kaplan–Meier method, and the log-rank test was used to compare overall survival (OS). Multivariable Cox proportional hazards models identified independent prognostic factors.

Results

The median patient age was 54 years (95% CI: 53–58), with 27.8% under 45 years. Tumours were located in the colon (42.0%) and rectum (41.6%), with adenocarcinomas comprising 89.0% of cases. Stage III and IV diagnoses accounted for 39.1% and 29.9% of cases, respectively. There were 38.8% deaths, with a median OS of 45 months (95% CI: 32.5–57.5) and a 5-year OS rate of 39.3% (95% CI: 28.7–49.9). Poor OS was independently associated with stage II [aHR = 13.50, (95% CI: 1.74–104.75)], stage IV [aHR = 10.41, (95% CI: 1.16–93.11)], smoking [aHR = 1.72, (95% CI: 1.06–2.81)], multiple comorbidities [aHR = 2.72, (95% CI: 1.15–6.46)], and treatment with herbal medicine [aHR = 1.58, (95% CI: 1.04–2.38)]. Working in the informal sector was linked to improved OS [aHR = 0.57, (95% CI: 0.36–0.89)].

Conclusions

CRC survival rates in this cohort were low, with a substantial proportion of patients experiencing mortality within five years following diagnosis. The findings highlight the critical need for early detection and intervention, particularly in younger populations to improve survival outcomes.

Introduction

Colorectal cancer (CRC) is the most commonly diagnosed malignancy in men and the second in women globally [1–3]. It is also the second leading cause of cancer-related mortality globally, following lung cancer [4]. Progress in scientific knowledge, treatment modalities, and medical screening programs have made significant contributions to improving the survival rates of patients with CRC [5, 6]. However, despite this progress, approximately 50% of CRC patients still do not survive beyond 5 years post-diagnosis [7]. Disparities in CRC survival, well-documented both regionally and globally [8, 9], highlight substantial variations in outcomes. In sub-Saharan Africa, for instance, 5-year overall survival (OS) rates range from 8% to 30% [10, 11], with studies in Ethiopia and South Africa reporting survival rates of 33% and 26%, respectively [10–13]. In contrast, high-income countries such as Israel report a 5-year OS of 72%, with rates between 65% and 70% across North America, Europe, and Australia [14–16].

Several risk factors contribute to CRC development, including genetic predisposition, high consumption of red meat smoking, obesity alcohol consumption and inflammatory bowel disease. Current prognosis traditionally depends on clinicopathological features, with key prognostic factors such as tumour site, grade, and size [17, 18]. Additionally, age, sex, family history, racial-ethnic disparities [19] significantly impact CRC mortality [20]. Genome-wide association studies have also identified poor prognostic factors related to tumour anatomical sites bowel wall infiltration, lymph node metastases, and distant metastases [21].

CRC symptoms often include change in bowel habit, weight loss, rectal bleeding, abdominal pain, intestinal obstruction, rectal mass and abdominal mass. In Ghana, limited awareness and socioeconomic barriers contribute to late-stage diagnosis, further worsening survival outcomes. Improved survival rates in advanced countries is attributed to early detection methods including colonoscopies, sigmoidoscopies, computed tomography colonography, and faecal blood testing [22]. Conversely, in many countries in sub-Saharan African countries including, limited healthcare infrastructure and access to curative treatments result in late-stage presentation, with an alarming 60% diagnosed at stage IV CRC and 5-year survival rate below 1% [11, 23].

CRC incidence is rising in sub-Saharan Africa, but comprehensive data on incidence, prevalence, and mortality are often lacking, leading to extrapolations from a few countries [24]. Recent evidence reveals increased CRC rates in African countries previously considered low-risk [25, 26]. In Ghana, for example, new CRC cases have surged eightfold annually since 1960, with this trend expected to continue due to population aging, increased life expectancy, and urbanization. CRC is the 10th most common cancer overall in Ghana. There is a near-equal gender prevalence; for men it’s the 5th and for women it’s the 7th and over half of the patients present with advanced disease. The Ghana National Cancer Steering Committee introduced guidelines recommending faecal occult blood testing for CRC screening to facilitate early detection in 2011. However, the implementation of these guidelines has been limited [27]. A previous analysis of CRC cases in Ghana (2009–2015) revealed that most patients were diagnosed at advanced stages, with a 5-year survival rate of just 16% [28]. Notably, survival decreased drastically from 90% in stage I to 0% in stage IV. Treatment approaches for CRC vary depending on disease stage and include snuggery chemotherapy, radiotherapy and targeted therapies. Treatment availability is often constrained by resource limitation, with target therapy being expensive and not covered by insurance in Ghana, Moreover, radiotherapy is only available in only three centres in Ghana [28]. These limitations undergird the critical need for improve screening, early diagnosis and treatment process.

This study aims to update CRC survival estimates and identify key prognostic factors of CRC in the Ghanaian context. The provision of insights into the current survival trends and determinants through these findings can inform policy initiatives and interventions aimed at improving CRC management and patient outcome in Ghana.

Methodology

Study design/setting

This study was a retrospective cohort investigation involving patients diagnosed with CRC at the Surgical and Oncological Department of the Komfo Anokye Teaching Hospital (KATH). KATH, located in Kumasi, the capital of Ashanti region, is the second-largest teaching hospital in Ghana and serves as a referral centre. Ashanti region had an estimated total population of 4,780,380 as of 2021 data from the Ghana Statistical Service.KATH boasts highly skilled medical staff including six Oncology specialist doctors, five oncology nurses, anaesthetists, and specialized departments like surgery, internal medicine, obstetrics, gynaecology, paediatrics, oncology, family medicine, and emergency care. The oncology directorate have radiotherapy and magnetic resonance imaging machines. Treatment regimens available includes chemotherapy, targeted therapy and hormonal therapies. With a thousand-bed capacity, the hospital’s strategic location within Kumasi, coupled with a well-connected road network in the region ensures accessibility for residents of the Ashanti Region and even those from more distant areas.

Study population and subject selection

We retrospectively retrieved all clinical CRC cases between 2017 and 2022. These cases were obtained from the medical records stored in the database of the Surgical and Oncology (S & O) Department. The final dataset constituted all the eligible patients within this period, totalling 281 individuals who met the predefined inclusion criteria. We did not perform prospective sample size or power calculation because the study utilised the entire available cases over the specified duration. Survival rates were reported with corresponding 95% confidence to enable the detection of clinically significant differences.

Inclusion criteria

Patients with complete clinical information and pathological assessment confirming malignancy in the large bowel were included.

Exclusion criteria

Those with different large bowel diseases and histopathologically confirmed non-malignant tumours were not considered for inclusion. Patients with incomplete data were also excluded from our analysis.

Variables

The information retrieved constituted various aspects including socio-demographic characteristics such as age, sex, occupation, and marital status. It also covers clinical information like duration of disease symptoms which was defined as estimated duration of symptoms before diagnosis, abdominal pains, blood stools, constipation and nausea, smoking and alcohol intake history, family history, presence of comorbidities (diabetes and hypertension), and tumour histopathological analysis such as tumour sites, histological type, tumour grade, TNM tumour staging, depth of tumour invasion(T1-T4), lymph node involvement(N0-N ≥ 2) and distant metastases (M0, M1). Additionally, data on treatment types including surgery, radiotherapy, and chemotherapy were also reviewed. Furthermore, body mass index (BMI) was calculated based on their recorded weight and height at the time of diagnosis.

Follow-up of study participants

Follow-up of patients was conducted using a combination of active or passive approaches. During their scheduled hospital follow-up appointments, patients received personalised outreach, while those unable to attend in-person were promptly connected through alternative means of communication by phone calls to family members. Deaths of subjects were confirmed by referencing clinical case records or via contact with their families and relatives. The duration of survival was calculated from the date of initial diagnosis to either the date of death or the last follow-up. Patients who remained alive at the end of the follow-up and those lost to follow-up were subjected to censoring, which occurred at the time of their last recorded contact (Fig. 1).

Fig. 1.

Workflow for the review and selection of patients

Statistical analysis

Data collected in the study was entered into Microsoft Excel 2021. Statistical analyses were performed on R language for statistical computing (Version 4.3.1) [29]. Patients with missing data for more than three variables analyzed in each category were excluded from the analysis. Descriptive analysis was done using measures such as frequency, percentage and median. Survival time was determined using the Kaplan–Meier method. The Log-rank test was applied to compare overall survival among various groups based on socio-demographic clinical and tumour characteristics. Cox regression models were fitted to identify prognostic factors associated with patient survival. The Proportional hazard (PH) assumptions were initially assessed for deviation using the scaled Schoenfeld residuals and global Schoenfeld test (GST). Notably, GST did not reveal statistical significance for the models (P >0.05). Multicollinearity was assessed before building multivariable models and the variation inflation factor value was obtained for covariates range of 1–4 indicating an absence of multicollinearity. Multivariate analyses were conducted using a force entry approach. The results of these models were presented in terms of adjusted hazard ratios (aHRs) with their corresponding 95% confidence intervals (CIs). P value < 0.05 was considered statistically significant.

Results

A total of 281 eligible patients were included in the analysis. The median age of patients was 54.5 years (SD = 15.3), with 27.8% of the cohort aged under 45 years. There was a slight male predominance (50.9% male vs. 49.1% female). Most patients were married (62.6%) and employed in the informal sector (62.6%). A substantial majority reported no family history of CRC (83.6%), no smoking history (82.2%), and no alcohol consumption (86.1%). Diabetes was present in 27.8% of the cohort, while 37.0% had hypertension. Only 14.9% of patients reported multiple comorbidities. The Log-rank test revealed no statistically significant differences in survival outcomes based on age (p = 0.543), sex (p = 0.627), marital status (p = 0.686), occupational status (p = 0.135), family history of CRC (p = 0.694), alcohol consumption (p = 0.698), diabetes (p = 0.694), or hypertension (p = 0.905). However, patients with a smoking history exhibited significantly lower survival rates (p = 0.024). Additionally, the presence of multiple comorbidities was associated with a trend toward reduced survival, although this did not reach statistical significance (p = 0.091) (Table 1).

Table 1.

Distribution median survival rates and Log-rank test for socio-demographic and lifestyle characteristics with OS

Variables	Frequency (%)	Median OS (95% CI) (months)	P-Value
Age (years)median (95% CI)	54.0(53.0–58.0)
Age Groups (years)			0.543
< 45	78 (27.8)	52.0 (27.6–76.4)
46–55	70 (24.9)	45.0 (12.0–77.9)
56–65*	63 (22.4)	-
≥ 65	70 (24.9)	39.0 (16.1–61.9)
Sex			0.627
Female	138 (49.1)	50.0 (38.3–61.7)
Male	143 (50.9)	37.0 (28.1–45.9)
Marital status			0.686
Divorced	23 (8.2)	34.0 (1.6–66.4)
Married	176 (62.6)	45.0 (27.2–62.8)
Single	25 (8.9)	62.0 (0.0–124.8)
Widowed	12 (4.3)	45.0 (32.5–57.8)
Occupational status			0.135
Formal	68 (24.2)	19.0 (6.2–31.8)
Informal	176 (62.6)	52.0 (37.9–66.0)
Retired	25 (8.9)	41.0 (7.0–74.9)
Student	12 (4.3)	38.0 (6.0–69.9)
Family history			0.694
No	235 (83.6)	47.0 (29.6–64.4)
Yes	40 (16.4)	38.0 (31.7–44.3)
Smoking history			0.024
No	231 (82.2)	50.0 (36.0–63.9)
Yes	50 (17.8)	19 (12.3–25.7)
Alcohol intake			0.498
No	242 (86.1)	60.0 (39.3–82.1)
Yes	39 (13.9)	41.0 (24.7–57.2)
Presence of comorbidities			0.091
No	239 (85.1)	50.0 (36.2–63.8)
Yes	42 (14.9)	19.0 (9.7–28.2)
Diabetes			0.694
No	203 (72.2)	45.0 (31.2–58.8)
Yes	78 (27.8)	47.0 (11.6–82.4)
Hypertension			0.905
No	177 (63.0)	45.0 (25.2–64.9)
Yes	104 (37.0)	50.0 (35.8–64.2)

SD Standard deviation, OS Overall survival, CI Confidence Interval

*Median survival time not achieved

As depicted in Table 2, most of the patients (55.2%) experienced symptoms within less than 6 months. Among the 129(45.9%) who underwent surgery as treatment, 89.1% had elective surgery while 10.9 required emergency surgery. One hundred and twenty-two patients (43.4%) received chemotherapy, 64(22.8%) received radiotherapy, and 74(26.3) engaged in herbal medicine treatment. The Log-rank test results revealed no statistically significant differences in survival outcomes related to the duration of symptoms (p = 0.410), surgery (p = 0.120), chemotherapy (p = 0.681), and radiotherapy (p = 0.164). Patients who opted for herbal medicine treatment experienced significantly shorter survival times compared to those who did not (26 months vs. 50 months, p = 0.019).

Table 2.

Distribution, median survival rates, and Log-rank test for clinical parameters with overall survival

Variables	Frequency (%)	Median OS (95% CI) (months)	P-Value
Duration of symptoms (months)			0.410
< 6	155 (55.2)	35.0 (19.8–50.2)
6–12	113 (40.2)	62.0 (36.4–87.6)
> 12	13 (4.6)	52.0 (0.0–129.2)
Surgery			0.120
No	152 (54.1)	29.0 (12.0–35.9)
Yes	129 (45.9)	60.0 (46.9–73.0)
Nature of surgery			0.115
Elective	115 (89.1)	60.0 (37.6–82.4)
Emergency*	14 (10.9)	-
Chemotherapy			0.681
No	159 (56.6)	45.0 (31.9–58.0)
Yes	122 (43.4)	37.0 (17.3–56.7)
Herbal medicine			0.019
No	207 (73.7)	50.0 (33.7–66.3)
Yes	74 (26.3)	26.0 (10.9–41.4)
Radiotherapy			0.164
No	217 (77.2)	39.0 (21.5–56.5)
Yes	64 (22.8)	60.0 (39.3–80.6)

OS overall survival, CI confidence interval

*Median survival time not achieved

As shown in Table 3, the most prevalent tumour locations were in the colon (42.0%) and rectum (41.6%). Majority of the patients exhibited adenocarcinomas (89.0%), and the tumours were predominately moderately differentiated (55.2%) and at stage III (39.1%). The T3 classification was found in 39.1% of cases, with N1 lymph node involvement in (50.9%) and M0 absence of distant metastasis in 75.1%. Statistically significant differences were observed in tumour staging (p = 0.001) and the presence of distant metastasis (0.035). The median OS time for patients with stage I tumours was not reached while those with stage IV had shorter survival time compared to those with stage II and III tumours, respectively. Furthermore, CRC patients with distant metastasis had shorter survival time than those without (24 months vs. 60 months, p = 0.035).

Table 3.

Distribution, median survival rates and Log-rank test for tumour pathological parameters with overall survival

Variables	Frequency (%)	Median OS (95% CI) (months)	P-Value
Histological type			0.269
Adenocarcinoma	250 (89.0)	47.0 (31.9–62.0)
Others	31 (11.0)	35.0 (13.3–56.7)
Histological grade			0.652
Moderately differentiated	155 (55.2)	60.0 (38.8–81.20
Poorly differentiated	65 (23.1)	35.0 (15.2–54.7)
Undifferentiated	22 (7.8)	25.0 (0.8–49.9)
Well-differentiated	39 (13.9)	52.0 (19.8–84.1)
Tumour site			0.226
Anal*	8 (2.8)	-
Anorectum	26 (9.3)	60.0 (44.9–72.5)
Colon	118 (42.0)	26.0 (9.0–42.9)
Rectum	117 (41.6)	50.0 (28.4–71.6)
More than one site*	12 (4.3)	-
Pathological stage			0.001
Stage I*	17 (6.1)	-
Stage II	70 (24.9)	35.0 (1.9–68.0)
Stage III	110 (39.1)	52.0 (34.2–69.7)
Stage IV	84 (29.9)	24.0 (13.1–34.8)
Depth of tumour invasion
T1*	13 (4.6)	-	0.264
T2	66 (23.5)	45.0 (35.8–57.4)
T3	110 (39.1)	50.0 (37.1–62.4)
T4	92 (32.7)	27.0 (10.8–43.2)
Lymph node metastasis			0.412
N0	61 (21.7)	62.0(27.0–96.9)
N1	143 (50.9)	39.0 (21.6–56.4)
≥N2	77 (27.4)	38.0 (18.2–57.8)
Distant metastasis			0.035
M0	211(75.1)	60.0 (45.2–74.8)
M1	70 (24.9)	24.0 (11.8–36.1)

OS Overall survival, CI Confidence interval, T Tumour depth, N Lymph node, M Distant metastasis

* Median survival time was not achieved

Figure 2 illustrates a survival function plot depicting the overall survival probability of CRC patients from 2016 to 2022. The plot indicates that as the observation time increases, the survival of patients decreases. Specifically, the survival rates at 1 st, 2nd, 3rd, 4th, and 5th years were 72.7% (95% CI: 68.9–76.5), 61.3% (95% CI: 55.5–67.1), 53.9% (95% CI: 46.7–61.1), 48.5% (95% CI: 40.4–56.6) and 39.3% (95% CI: 28.7–49.9) respectively. The median overall survival time was determined to be 45 months (95% CI: 32.5–57.5) (Fig. 2).

Fig. 2.

Overall survival probability of CRC patients

Figure 3 shows that the survival rate was among patients with different stages of tumours. The median OS time for patients with stage I tumours was not reached while those with stage IV (24.0 months, 95%CI: 13.1–34.8) had shorter survival time compared to those with stage II (35.0 months, 95% CI:1.9–68.0) and III tumours (52 months, 95% CI: 34.2–69.7), respectively. Patients with stage I tumours showed a high 5-year survival rate of 93.8% (95% CI: 92.0–95.6). In contrast, those with stage II had a lower 5-year survival rate of 37.5% (95% CI: 30.7–44.3), while patients with stage III showed a 5-year survival rate of 41.5% (95% CI: 36.1–47.5) and those with stage IV had 5-years survival rate of 16.4% (95% CI: 6.9–25.9). These differences in survival rates across the different cancer stages were found to be statistically significant (p = 0.0001) (Fig. 3).

Fig. 3.

Kaplan-Meier curves showing CRC-specific survival rates based on tumour stage.

Figure 4 presents the result of a Cox proportional hazard regression analysis aimed to determine aHRs for various socio-demographic and lifestyle characteristics factors on CRC patients’ survival. The multivariate analysis revealed that the presence of multiple comorbidities (p = 0.023), a history of smoking (p = 0.029), and employment in the informal sector were identified as significant determinants of patients’ survival. The risk of CRC-related death was 2.72 times higher in patients with multiple comorbidities compared to those without such conditions. Smoking history was significantly associated with worse OS with an HR of 1.72 (95% CI: 1.06–2.81). Conversely, patients working in the informal sector experienced an improved OS with an HR of 0.57 (95% CI: 0.36–0.89) (Fig. 4).

Fig. 4.

Overall survival by socio-demographic and lifestyle characteristics status: multivariate analysis of prognostic factors. aHR: adjusted hazard ratio, CI: confidence interval, AIC: Akaike information criterion

Figure 5 shows the association between clinical parameters and OS through Cox regression analysis. In a multivariate analysis, the result indicated that patients’ engagement in herbal medicine treatment modality had reduced OS with an adjusted HR of 1.58 (95% CI: 1.04–2.38, p = 0.031). Although not statistically significant, it is worth noting that surgery [HR = 0.79 (95% CI: 0.53–1.16), P = 0.223] and chemotherapy [HR = 0.96 (95% CI: 0.65–1.42), P = 0.837] treatment modalities appeared to be associated with improved OS. In contrast, being overweight was tentatively linked with poorer OS [HR = 1.27 (95% CI: 0.75–2.15, P = 0.373) (Fig. 5).

Fig. 5.

Overall survival by clinical parameters status: multivariate analysis of prognostic factors. aHR: adjusted hazard ratio, CI: confidence interval, AIC: Akaike information criterion

Figure 6 shows the association between tumour pathological factors and OS. The patient’s tumour stage at the time of diagnosis was found to be significantly associated with worse OS. Specifically, being diagnosed at stage IV [HR = 10.41 (95% CI: 1.16–93.11), P = 0.036], as well as stage II [HR = 13.50 (95% CI: 1.74–104.75), P = 0.013] were significantly both associated with lower survival rates according to multivariate analysis. The presence of distant metastasis was significantly linked to reduced OS in the univariate analysis, [HR = 1.53(95% CI: 1.02–2.33), P = 0.038]. However, this association was not statistically significant in the multivariate analysis [HR = 1.29 (95% CI: 0.78–2.16), P = 0.323]. With regard to tumour locations, depth of invasion, and lymph node metastasis, while they appeared to have a reduced effect on OS, these factors were not statistically significant (P > 0.06) (Fig. 6).

Fig. 6.

Overall survival by tumour pathological characteristics: multivariate analysis of prognostic factors. aHR: adjusted hazard ratio, CI: confidence interval, AIC: Akaike information criterion

Discussion

Numerous studies have documented the increasing incidence of colorectal cancer (CRC) in low- and middle-income countries (LMICs), with limited access to early detection and comprehensive treatment negatively affecting survival outcomes. This study updates survival rates and identifies prognostic determinants of CRC in a Ghanaian cohort receiving care at KATH. CRC has been predominantly associated with older age groups, and routine screening typically begins at age 50 years for those at average risk. Previous studies in Sub-Saharan Africa have reported a median age at diagnosis of approximately 54.9 to 56 years [30, 31] However, recent studies in Europe [32, 33] and America [34, 35] have reported a concerning rise in colorectal cancer cases among younger patients, especially those under 45 years old. This trend is consistent with the findings of our study, where a significant portion of the patients diagnosed with CRC were below the age of 45 years. As a result, CRC screening guidelines in many countries, including the United States, have recently been revised to begin at 45 years, but younger patients in Ghana may still be missed by traditional screening programs, contributing to later-stage diagnoses and poorer outcomes. This upward trend may be attributed to changes in dietary habits, such as a rise in processed foods and a decline in fibre intake among the young population in Ghana [36]. Additionally, the subtlety of CRC symptoms in younger individuals, combined with delayed medical attention, exacerbates this issue [37].

The present builds upon earlier work on CRC survival rates in Ghana which reported a 5-year survival rate of 16% between 2009 and 2015, the lowest among globally [25]. In the present the 5-year OS rate has improved to 39.3%, indicating progress in patient outcomes. However, these rates still lag significantly behind those in advanced countries such as the United States (64.9%), the United Kingdom (60.0%), Japan (67.8%), and South Korea (74.3%) [38, 39]. Additionally, our reported CRC survival rates is lower than that of Mauritius (78.4%), Seychelles (59.1%), and Namibia (43.8%) [40]. Comparative analyses from other sub-Saharan African nations also highlights, persistent survival disparities. Joko-Fru et al. reported a 5-year OS rates of 36.8% in Ethiopia, 34.8% in South Africa, 30.3% in Cote d’Ivoire, and 30.2% in Uganda, all of which remain lower that the OS rate observed in this study (Joko-Fru, 2024). Hassen et al. further documented an OS 5-year CRC survival rate of 0.28 (95% CI, 0.19–0.38) in sub-Saharan Africa, with middle-income countries (0.31; 95% CI: 0.17–0.49) demonstrating a slightly higher survival rate compared to low-income countries (0.20; 95% CI: 0.11–0.35) [11]. A separate study by Teka et al., in Ethiopia, reported an even lower survival rate 28.7% [41].

This modest improvement in CRC survival in this study compared to our previous study [25] may, in part be, attributed to the younger age distribution of the cohort, some degree of CRC screening implementation, and improvement in cancer care provision including gradual improvement in surgical interventions, and possible changes in treatment approaches [42]. However, the effective rollout of screening programs remains a critical challenge. Previous studies have shown substantial variability in physician adherence to screening guidelines, with only a minority of physicians adhering to national recommendations [43]. Furthermore, the absence of modern cancer care infrastructure and a comprehensive national cancer strategy continues to hinder survival outcomes in Ghana [44].

In the majority of cases, the stage at which cancer is initially detected has a significant impact on an individual’s survival prospects [45]. In our study, patients with stage I tumours showed a high survival rate of 93.8%. However, those with stage II had a low 5-year survival rate of 37.5%, while patients with stage III showed a survival rate of 41.5% and those with stage IV had a survival rate of 16.4% These differences in survival rates across the different cancer stages were found to be statistically significant. Further, exploratory analysis showed that half of the stage II patient who did not survive also had a history of hypertension, potentially contributing to their reduced survival time. Several groups attributed the survival paradox to either stage migration or lack of receipt of systemic therapy [46, 47]. Moreover, retrieved pathological reports were from different laboratories, suggesting that tumours were staged by different pathologists, raising the possibility that some tumours may have been under-staged, contributing to the unfavourable prognosis in stage II disease. It is noteworthy that the 5-year survival rates for colorectal cancer for all stages, in our study were comparatively lower than those reported in developed countries [44, 45]. Yijun Li et al., [48] reported that patients with CRC had the following five-year survival rates: 91.0% for stage IIIA, 87.6% for stages IIA, and 72.9% for IIIB in the United States. Patients in stages IIB and IIC demonstrated less favourable cancer-specific survival rates, with 5-year survival rates of 68.9% and 65.6%, respectively. Al-Ahwal and colleagues [49], in their study conducted in Saudi Arabia, reported that patients with stage I cancers had a survival rate of 63.3%, while those with stages II and III CRC had a survival rate of 50.2%. However, for patients diagnosed with stage IV cancers, the survival rate was notably lower at 14.7%. Overall, our results show an improvement in CRC compared to the previous study by Agyemang-Yeboah et al. [25] who reported overall survival of 90% for stage I, 34% for stage II, 12% for stage III and 0% for stage IV. The improvement in treatment outcome could be attributed, in part, to the early detection of CRC in select cases and improvement in treatment modalities. Furthermore, our results demonstrated that both T3-T4 and progressive increase in nodal involvement (N1, N2) were associated with increased risk, although statistically significance was not reached. Previous studies have shown that tumours at T3-T4 stage and a substantial degree of nodal involvement (N2) are considered as high-risk in CRC patients [50, 51].

Another interesting finding in our study is the age distribution at presentation. Patients who were younger than 45 years old constituted 27.8% of the patients while patients younger than 55 years old made up more than 52% of the study population. This is in contrast with what is found in the United States of America where patients younger than 50 years make up 13% of the total CRC population while patients older than 65 years constitute 56% of cases [52].

Many studies have investigated the association of smoking status with CRC-specific survival [53, 54]. While results from these studies have yielded varying outcomes, they, on the whole, suggest a stronger association between smoking and poorer survival among CRC patients. In a recent study conducted by Alwers et al. [55], it was observed that individuals with a greater history of smoking in pack years were significantly 1.4 times at risk of experiencing poorer CRC-specific OS. In concordance with these findings, our results also revealed a significant association between smoking history and inferior OS. Furthermore, nearly every study that had explored the link between cigarette smoking and adenomatous polyps has consistently reported an elevated risk [2, 56] Most recently, research interest has focused on the possibility that risk associated with smoking is influenced by genetic variations in metabolizing enzymes [57]. Patients who had multiple comorbidities had a 2.72-fold elevated likelihood of dying from CRC compared to those without such health issues. These observed results concur with reports from a prior study conducted by Luque-Fernandez and colleagues [34]. In their study, the authors found that individuals with multiple comorbidities had a twofold higher risk of mortality compared to those without comorbidities. Many other studies have also documented that patients with diabetes and hypertension who develop CRC are more likely to die early compared to those without diabetes and hypertension thereby corroborating with the finding of our study [58, 59].

Surgery and chemotherapy have traditionally been the primary interventions for CRC patients, but the prognosis, especially, for those with metastatic lesions is unsatisfactory [60, 61]. Although statistical significance was not achieved in this study, both treatments showed a positive trend towards OS. Agyemang-Yeboah et al. [25], reported a significant association between chemotherapy and improved survival in CRC in a Ghanaian population. Moreover, the administration of chemotherapy to patients’ post-surgery enhances prognosis by eliminating minuscule cancer cells or molecular residual diseases that can progress into more substantial tumours in the future. Adjuvant chemotherapy for stage III CRC has been showed to increase survival rates by 15–20% [62]. Moreover, a study conducted by [63] in the United States showed an improved and clinically relevant increase in OS from 20% to 33% after 5 years regardless of high-risk tumour pathologic features, when adjuvant chemotherapy was administrated across all patient subgroups. Given that most CRC patients in this study were in the younger population and primarily in stages I-III, improved access to surgery and oncology care can significantly benefit many early-stage CRC patients.

The results of the study revealed that herbal medicine use was significantly associated with reduced OS. This is in contrast to a study by Nurumal et al. [64] where positive results on traditional Chinese and Japanese herbal treatments for CRC which included lower tumour markers, improved post-surgery gastrointestinal function, enhanced disease control rate, a better quality of life score and decreased oxaliplatin-induced peripheral neurotoxicity. Moreover, Androulakis et al. [65] identified Khaya senegalensis bark extract, popular in West Africa, as a potential natural CRC chemo-preventive agent in human CRC lines. Smit et al. [66] also suggested anticancer potential in S. frutescens and X. undulatum via in vitro tests, requiring further clinical investigation. The discrepancies may stem from self-prescribed, potentially untested herbal medicines in our CRC cohort.

The current study has some limitations. It was conducted in a single referral hospital in Ghana’s middle belt, potentially not fully reflective of the entire Ghana. These numbers are a gross underestimate of the true prevalence of CRC in Ghana as the majority of patients never make it to the hospital and may die at home. Some prognostic indices such as perineural invasion, pre-and-post operative CEA levels and treatment complications were absent in the data due to their secondary nature. Additionally, while the death of 26 CRC cases was confirmed through medical records 83 others were identified via telephone interview, potentially introducing bias in death estimation. Additionally, the current study provides updated survival estimates and allows for comparison with our previous findings, however a prospective longitudinal study would have been enabled a more detailed analysis of treatment effects, disease progression and patient outcome over time. Furthermore, molecular tumours analysis including MMR status, RAS and BRAF were not available. Despite limitations, the study provides valuable insights into CRC survival in Ghana, providing survival estimates and identification of key influencing factors, that contribute to a larger puzzle of understanding CRC dynamics globally. Future multi-center prospective studies could offer a broader perspective.

Conclusion

CRC survival rates in this cohort were low, with more than half of patients experiencing mortality within five years following diagnosis. The findings highlight the critical need for early detection strategies, improved access to treatment and strengthened healthcare infrastructure intervention, particularly for younger populations to improve survival outcomes. Future research that focuses on prospective, multi-centre studies to provide a more comprehensive understanding of CRC survival trends is warranted.

Abbreviations

aHR

Adjusted Hazard Ratio

CI

Confidence Interval

CRC

Colorectal Cancer

GST

Global Schoenfeld Test

KATH

Komfo Anokye Teaching Hospital

SD

Standard deviation

OS

Overall survival

Authors’ contributions

TB, SS, EA, EOB, CO, FAY and JY participated in Conceptualization, Investigation and Methodology. TB, EOB, DS, EB, ES, MN and FA participated in Data curation and TB, BMN, JL, MN and EA, participated in Formal analysis and writing, reviewing, and editing. All authors read and approved the final manuscript.

Funding

No funding was obtained for this study.

Data availability

The datasets during and/or analysed during the current study are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

Ethical Approval for the study was obtained from the Institutional Review Board, Komfo Anokye Teaching Hospital (KATH/IRB/AP/156/22). Written informed consent was obtained from all participants and documented from their medical records. This study was conducted in accordance with the core principles outlined in the Declaration of Helsinki.

Consent of publication

Not applicable.

Competing interests

The authors declare no competing interests.