Correlation between total prostate specific antigen and histological grading of prostate cancer in Kenyan mission hospital: a five-year retrospective review

Abstract

Background

Prostate cancer (CaP) is the leading non-cutaneous cancer in men of African descent, with the higher mortality rates found in sub-Saharan Africa. Early diagnosis, staging, and management of prostate cancer could help curb its mortality rate in SSA. However, access to precise radiological imaging for staging purposes is limited in our setting. We sought to evaluate the correlation between total prostate specific antigen (tPSA) and histological grading of CaP in our resource-limited setting.

Method

We conducted a retrospective review of records of patients treated for biopsy-proven CaP at the AICKH diagnosed between January 2018 and December 2022. We excluded patients who were already on any sort of treatment of bladder outlet obstruction and incomplete charts. We used Spearman correlation coefficients, and ANOVA to evaluate the relationship between tPSA and various grading parameters. A P-value less than 0.05 was considered significant.

Results

We included 327 medical records. The mean tPSA was 112 ± 4.5ng/ml. The most common Gleason score and grade group were 8 (33.8%) and 4 (33.8%) respectively. Perineural involvement was present in 33% of our population. The tPSA had a positive correlation with Gleason score (rho = 0.253, p < 0.001), grade group (rho = 0.296, p < 0.001), perineural involvement (rho = 0.241, p = 0.001) and proportion of sample invasion (rho = 0.171, p = 0.005). A linear and homogenous variance existed in mean tPSA across increasing Gleason score (p < 0.001).

Conclusion

tPSA is a good predictor of the severity of CaP in resource-limited settings and can be used to inform management decisions.

Introduction

Prostate cancer (CaP) is a significant global health concern, ranking as the third most diagnosed malignancy worldwide, with an estimated 1,414,259 new cases in 2020 and its accounts for 7.3% of all cancer cases globally [1]. CaP particularly affect populations in high Human Development Index (HDI) nations where incidence rates reach 37.5 per 100,000 compared to 11.3 per 100,000 in low-HDI countries [2]. CaP in sub-Saharan Africa (SSA) reports indicate some of the highest mortality rates worldwide [1, 3]. CaP incidence in SSA may be underreported and could match or exceed rates in African American men [4, 5]. The aggressive nature of CaP in SSA populations, combined with late stage diagnoses, contributes significantly to poor outcomes [6]. Barriers to accurate incidence estimation include limited infrastructure for population-based cancer registration and restricted access to prostate-specific antigen (PSA) testing [7].

PSA is a crucial biomarker for prostate cancer detection and monitoring, with serum levels reflecting tumor burden [8]. The introduction of PSA testing in the late 1980s significantly improved early detection rates and led to shifts in treatment strategies [9]. However, the specificity of PSA testing remains a limitation, as elevated levels can result from benign prostatic hyperplasia or prostatitis in addition to malignancy [10]. Despite these limitations, PSA testing remains widely used in high-income countries, where it is complemented by advanced imaging modalities such as multiparametric magnetic resonance imaging (MRI) to enhance diagnostic and staging accuracy [11].

In SSA, including Kenya, access to advanced imaging for staging remains largely inaccessible [12]. Consequently, clinical decision-making often relies heavily on histopathological grading, particularly the Gleason score, which is used to determine tumor aggressiveness and guide treatment strategies [13]. While PSA levels are widely used as a surrogate marker for disease severity, limited studies have examined the correlation between total PSA (tPSA) and histopathological grading in resource-limited settings [14]. Understanding this correlation is essential, as it may provide a cost-effective alternative for assessing disease severity in settings where MRI and other staging tools are inaccessible. This study examines the correlation between tPSA levels and prostate cancer histological grading at AIC Kijabe Hospital, Kenya, to assess tPSA as a predictor of disease severity in a resource-limited setting.

Methods

Study design and setting

We conducted a hospital-based retrospective review at the Urology Department of AICKH, a hospital with a bed capacity of 300. The laboratory could perform tPSA assays with quantitative results but not free PSA. Three pathologists provided prompt histological results, and the oncology department included a multidisciplinary team capable of handling advanced prostate cancer care.

Study period

January 1, 2018 to December 31, 2022.

Source of population

Records of patients treated for prostate cancer in the Urology Department of the AICKH.

Inclusion criteria

Records of patients treated for biopsy-proven prostate cancer at the AICKH whose biopsy was done at this institution.

Exclusion criteria:

• i. Incomplete medical records, that is, records that did not have PSA, pathological diagnosis, or that did not have all the cancer-grading information.

• ii. Patients who were on Finasteride, alpha-blocker, or a combination of both for more than one month before tPSA was done as some studies have revealed they influence PSA levels even though finasteride more than alpha blockers.

• iii. Patients whose tPSA was done more than two months before biopsy.

• iv. Patients with urinary tract infections as these can influence PSA levels.

Data collection and statistical analysis

We checked patients who had histopathology confirmed CaP in AICKH during the study period from the pathology lab database. We then used this information, went through the EMR (Electronic medical record), and patients who met the inclusion criteria and had no exclusion criteria were included in the study. We collected data using predesigned data collection sheets, which were tested in the same facility and upgraded accordingly. We trained three people to help the principal investigator collect the required data. Data collected included sociodemographic profile, tPSA results before biopsy, and biopsy results.

Variables

Since this was a retrospective study, all variables were independent variables, including, psa, gleason score, gleason grade group, proportion of sample involved with cancer, lymphovascular involvement, and perineural involvement.

Data management and analysis

We checked the data collection sheets for completion and accuracy of collection. We input the data into an Excel spreadsheet on a computer with a password known only by the primary investigator.

Data analysis was done using STATA and Statistical Package for Social Sciences (SPSS) version 27.0. Our dependent variable was dichotomous, cancer, or benign findings. Continuous variables were interpreted in terms of measures of central tendency, while categorical variables were expressed as proportions. ANOVA test was used to identify the difference in means between groups of continuous variables with more than two groups, while t-test was used for two grouped continuous variables. Spearman correlation analysis was done to get correlations between dependent and independent variables. For this analysis, continuous variable was the tPSA while gleason score, grade group, proportion of sample invasion were all categorized.

Results

A review of complete medical records of 710 patients treated for prostate disease during this study period revealed 327 with prostate cancer who met the inclusion criteria for the study. The mean age of the participants was 73.5 ± 9.5 years. Most patients (n = 178, 54.4%) were between 50 and 75 years old, and just four cases of prostate cancer were less than 50 years old.

The most common Gleason’s Score was 8 (n = 108, 33.8%). The most common grade group was Category 4 (n = 108, 33.8%). One-third (n = 101, 33.3%) of the patients had peri-neural involvement. Less than 5% (n = 13, 4.2%) of the patients had lymphovascular involvement. All of our patients had adenocarcinoma of the prostate. Analysis of the proportion of sample invaded by cancer was recorded in 269 patients. The mean proportion of sample invasion was 59.9 ± 1.92%, and the median invasion was 36.2%. More than half of the patients had cancer involvement above 50% (n = 162), and more than a third had invasion above 75% (n = 98). (Table 1).

Table 1 The histopathological features of prostate cancer in AICKH

Generally, the mean PSA value was 112.6 ± 4.5 ng/ml. Patients with PSA values between 20ng/ml made up 90.5% (n = 296) of the total number of patients, 48.6% (n = 144) of which had PSA values greater than 100 ng/ml. A few patients had PSA levels below 10ng/ml. (Table 2).

Table 2 Distribution of PSA values amongst patients with prostate cancer

There was a positive correlation between PSA and Gleason score (rho = 0.253, p < 0.001), grade group (rho = 0.296, p < 0.001), the proportion of sample involved/invaded with cancer (rho = 0.171, p = 0.005), and perineural involvement (rho = 0.241, p = 0.001). There was no significant correlation between PSA and lympho-vascular cancer spread.

The mean difference in PSA value was significant with varying Gleason score (F = 8.44, p < 0.001) and grade group (F = 12.43, p < 0.001) but was insignificant with the variation in proportions of the samples involved with cancer (F = 1.343, p = 0.261). The mean PSA of patients with perineural involvement was significantly higher than those without (134 vs. 102, p = 0.001). There was a linear and homogenous variance in the mean PSA across increasing Gleason scores (LS = 1.291, 0.274, p < 0.001), but the variations in the means associated with grade groups were unequal. Patients in grades 4 and 5 had significantly higher mean PSA values than grades 1–3(LS = 2.5, 0.042, p < 0.001). (Table 3).

Table 3 Test of homogeneity of variance amongst variable groups

Discussion

Prostate-specific antigen (PSA) is a biomarker commonly used to screen and monitor patients with prostate cancer [15]. Data on the use of tumor markers, such as serum PSA, for predicting prostate cancer histopathological staging, Gleason score, and lymphovascular and perineural involvement in patients with prostate cancer are scarce in sub-Saharan Africa.

We found a mean total serum PSA level of 112.6 ± 4.5 ng/ml. Most of the patients had PSA levels higher than 20ng/ml. Zivkovic and Pudasaini et al. reported similar results [16, 17]. Anunobi et al. mentioned that the majority of cancer patients in their study had a serum PSA level greater than 100 ng/ml. This study found that grade 4 tumors were the most common histopathologic grade group, a finding consistent with that of Pudasaini et al. and Anunobi et al. In contrast, other studies have reported grade 3 tumors to be more common [16, 18, 19].

The histopathology of prostate cancer in our study showed that 100% of the cases were adenocarcinoma. Pudasaini and colleagues found similar results [17]. Anunobi et al. also identified adenocarcinoma as the most common malignant prostate lesion, accounting for 99.1% of cases, with the remaining histologic types being squamous cell carcinoma and neuroendocrine tumors [20]. The glandular structure of the prostate explains the predominance of adenocarcinoma [19]; however, the absence of other histologic types in our study lacks a clear rationale.

Lymphovascular invasion, which reflects the presence of tumor cells in an endothelium-lined space, is a critical indicator of the severity of prostate cancer [21, 22]. In this study, lymphovascular invasion was identified in less than 5% of prostate cancer histopathology specimens. This figure is lower than the 10.2% and 12.4% reported by May et al. and De la Taille et al. respectively [21, 23]. The lower percentage in our study may be due to factors such as specimen collection and processing, sampling error, and the difficulty of assessing the pathologic stage in needle biopsy specimens [24, 25]. Larger specimens, such as those obtained during prostatectomy, may have a higher detection rate of lymphovascular invasion on histologic examination [26]. In addition, our study could not statistically demonstrate a correlation between serum PSA and lymphovascular invasion. Jung et al. reported that lymphovascular invasion was associated with the aggressiveness of prostate cancer, but failed to show an association between LVI and PSA [22]. De la Taille et al. found conflicting results in a study evaluating LVI in radical prostatectomy specimens [23]. A possible explanation could be the low percentage of lymphovascular invasion observed in our analyzed sample. This reason was also mentioned in the study by Jung et al [22]. In contrast, other studies have shown that preoperative serum PSA is a predictor of the presence of tumor cells in endothelium-lined spaces [27].

Preoperative PSA levels are also associated with perineural invasion, which was found in one-third of the study participants. This finding is consistent with other studies that highlight the significant correlation between PSA levels and perineural invasion, suggesting that PSA levels could serve as a predictive marker for determining whether a cancer patient will exhibit perineural invasion [22]. De la Taille et al. reported that perineural invasion and serum PSA are preoperative predictors of pathologic staging in patients who have undergone radical prostatectomy for malignancy [28]. This finding further underscores the utility of serum PSA for prostate cancer staging. For example, Szot and colleagues demonstrated a correlation between PSA levels and the presence of metastases on bone scans, which are critical for staging and treatment planning [29].

Our study demonstrated a correlation between total serum PSA and the histologic grade of prostate cancer. In addition, we found that serum PSA was a significant predictor of the Gleason score, which is used to assess the histological grade of the tumor. An increase in serum PSA has been associated with poorer differentiation of prostate cancer [16, 30, 31]. Several studies have concurred on a positive correlation between serum PSA and Gleason score [18, 32], which has implications for determining disease severity. However, the limitations of relying solely on PSA to predict pathologic stages on an individual basis have been recognized. Proposed reasons for this include the presence of confounding factors such as benign prostatic hyperplasia or inflammation within the gland, and the observation that PSA production decreases with increasing tumor volume in higher-grade lesions [33]. This trend was evident in our study, as the mean PSA level decreased after a Gleason score of 8 and between grade groups 4 and 5.

Limitations

First, our study was a hospital-based retrospective analysis, which introduces potential biases, such as challenges in identifying confounders and other related biases associated with this type of study design. Second, there are new parameters for evaluating patients with suspected prostate cancer. These parameters include free PSA, PSA ratio (the ratio of free PSA to total PSA), and PSA density (serum PSA divided by prostate volume as measured by ultrasound). Due to their specificity, these parameters are more effective predictors of the histopathologic grade of prostate cancer than tPSA [24]. A limitation of our study is the inability of most laboratories in the sub-Saharan region, including our study setting, to perform free PSA tests. In addition, the lack of advanced imaging such as magnetic resonance imaging in our setting hinders the proper staging of prostate cancer hence the need to search for other tools such as serum PSA in determining the grading and suggested stage of the disease. Finally, as a single institution study, it is likely to underestimate the true situation in Kenya and SSA. More studies without these limitations will be necessary to further emphasize our findings.

Conclusion

This study highlights the importance of tPSA in resource-limited settings for predicting the histopathologic severity of prostate cancer. It demonstrates a correlation between tPSA levels, histologic grade, Gleason score and perineural involvement. Consequently, this biomarker could be used to inform decision making.