Racial disparities in alpha‐fetoprotein testing and alpha‐fetoprotein status associated with the diagnosis and outcome of hepatocellular carcinoma patients

Abstract Background The use of alpha‐fetoprotein (AFP) testing for the surveillance, diagnosis, and prognosis of hepatocellular carcinoma (HCC) remains controversial. Here, we compared AFP testing rates, elevated AFP rates, factors associated with elevated AFP levels, and prognostic factors associated with overall survival (OS) in HCC patients from different ethnic groups. Methods Patients with HCC were identified from the Surveillance, Epidemiology, and End Results registries. Race was categorized as white, black, and others. AFP testing rates and elevated AFP rates were analyzed. Multivariable logistic regression and Cox regression analyses were used to identify independent factors associated with elevated AFP levels and prognosis, respectively. All statistical tests were two sided. Results A proportion of 79.2% of total HCC patients had AFP testing reports; 77.3% of white, 79.7% of black, and 81.2% of other races underwent AFP testing. Compared with white and other races, black HCC patients had a higher rate of elevated AFP levels among all patients and the early‐stage HCC patient cohort. Elevated AFP level was a significant prognostic factor for all HCC patients in different race groups. Factors associated with elevated AFP level and prognostic factors associated with OS varied significantly by race. Conclusions AFP testing, elevated AFP rates, predictors of elevated AFP level, and prognostic factors associated with OS differed significantly according to race after adjusting for AFP levels among the three groups. AFP testing for the surveillance, diagnosis, and prognosis of HCC patients is strongly recommended, although racial disparities need to be considered.


| INTRODUCTION
Hepatocellular carcinoma (HCC) is one of the most common cancers and the fourth leading cause of cancer-related mortality worldwide. 1 The incidence rate of HCC shows the fastest increase among all cancers in the United States, and the number of HCC cases have doubled in the last decade likely because of increased obesity/diabetes. 2 HCC mortality rates are increasing in the United States, although the burden of HCC is not equally distributed among different races. 3 Biannual screening of high-risk populations via ultrasonography is not widely adopted by physicians in the United States, despite the fact that early detection improves the prognosis of HCC patients. 4,5 The rationale for HCC screening in patients with high-risk factors (cirrhosis and hepatitis) is that screening tests such as ultrasonography or serum alphafetoprotein (AFP) measurement may help identify HCC patients at an early stage, at which the disease is potentially curative or has life-prolonging treatment options, including local destruction, surgical resection, or liver transplantation. 6 Curative treatments are only available for HCC patients diagnosed at an early stage, and the prognosis of HCC depends on tumor stage at diagnosis. 7 Early-stage HCC patients have 5year survival rates approaching 70% after surgical resection or liver transplantation, compared with a median survival of 1 year for those at advanced stages. 8 The incidence and mortality rates of HCC vary according to race/ethnicity, which is mainly attributed to differences in the prevalence of major risk factors and disparities in access to high-quality care. 9 Black race HCC patients are more likely to be diagnosed at advanced stages, less likely to follow the recommended treatment regimens, and more like to have worse survival outcomes compared with white race patients. 10 So, it is more important to perform the screening tests in high-risk black populations. Racial differences in the levels of AFP as a diagnostic marker for HCC have been investigated in different studies, and opposite results were reported. Some researchers consider that AFP is not accurate for the diagnosis of HCC in African Americans, 11 whereas other groups reported that African Americans show increased levels of AFP. 12 These conclusions were made based on data from multicenter clinical studies that included hundreds of patients with hepatitis C. However, to the best of our knowledge, no population-based studies have been performed to date.
In this report, we examined the rates of AFP testing performed and the proportions of elevated AFP levels in HCC patients using a United States population-based database, and reported contemporary overall survival (OS) rates for three different racial groups.

Results (SEER) data and selection of patients
SEER program data are maintained by the United States National Cancer Institute, including information on cancer incidence, pathology, and survival for approximately 28% of the US population. All data were obtained from the publicly available SEER database using SEER*Stat 8.3.5 (NCI). Patients presenting with HCC between 2004 and 2015 were identified using the International Classification of Diseases for Oncology, Third Edition (ICD-O-3) topography and morphology codes from SEER 18 Regs Research Data. Our preliminary sample included patients with "primary site" code "C22.0" and "histologic type" codes "8170 to 8175". Among 68 473 HCC cases identified, patients were excluded in a stepwise fashion if they had unknown age or age < 18 years old (n = 138), unknown race (n = 280), unknown AJCC TNM stage of cancer (n = 12 173), as well as autopsy only or death certificate only (n = 681) cases, leaving a final analysis sample of 55 201 HCC cases. Patients with available AFP testing records were selected for further analyses of the factors associated with elevated AFP levels at HCC diagnosis and survival. The dataset used is publicly available, and approval by the institutional review board at Nanjing Drum Tower Hospital (Nanjing, Jiangsu, China) was waived.
Conclusions: AFP testing, elevated AFP rates, predictors of elevated AFP level, and prognostic factors associated with OS differed significantly according to race after adjusting for AFP levels among the three groups. AFP testing for the surveillance, diagnosis, and prognosis of HCC patients is strongly recommended, although racial disparities need to be considered.

K E Y W O R D S
alpha-fetoprotein, diagnosis, end results, epidemiology, hepatocellular carcinoma, overall survival, racial disparities, surveillance

| Independent variables
The SEER database code data on the AFP level at diagnosis as "Positive/elevated," "Negative/normal," "Borderline; undetermined whether positive or negative," and other undocumented situations. HCC patients with AFP data classified as "Positive/ elevated," "Negative/normal," and "Borderline; undetermined whether positive or negative," were identified as "AFP testing performed". The SEER program uses the physician's determination of AFP levels reported in the medical record when available or the reference values provided by the laboratory reporting the result. We grouped HCC according to race as "whites," "blacks," and "others" (American Indians/Alaska Natives, and Asian/Pacific Islanders). Age was divided into quintiles ; and the stage of HCC was grouped into quartiles (stage I vs stage II vs stage III vs stage IV) following the 6th edition of the AJCC Cancer Staging Manual. The type of treatment was defined as "surgery performed," "Not recommended," and "Others."

| Statistical analysis
Descriptive statistics were reported as medians with interquartile range for continuous variables and as whole numbers and percentages for categorical variables. Descriptive statistics for discrete variables were compared using the Chi square test. Factors associated with AFP testing were identified using multivariable logistic regression analysis. The Kaplan-Meier method was used to assess OS stratified by race and APF level. A multivariable Cox proportional-hazards model was built to determine the prognostic factors of OS. The prognostic power of covariates was expressed as hazard ratio (HR) with 95% confidence interval (CI). Statistical significance was defined as P < .01 (two sided). All analyses were performed using SPSS version 23.0 (IBM).

| Patient characteristics
A total of 55 201 patients with HCC who were diagnosed between 2004 and 2015 and who met the study inclusion criteria were identified, including 33 789 whites (68.5%), 7628 blacks (13.8%), and 9784 others (17.7%). More than three quarters of patients (76.9%) were male. The median age of patients with HCC was 62 years (interquartile range, 56-71 years) and varied according to race (63 years for whites, 56 years for blacks, and 56 years for others). The number of HCC diagnosed increased from 9492 cases in the 2004-2006 period to 17 844 cases in 2013-2015. More than 79.2% of patients had AFP testing records. Other demographic and clinical characteristics of the cohort were shown in Table 1.

| AFP testing disparities stratified by race and diagnostic period
The overall proportion of patients who underwent AFP testing was higher in black (79.7%) and other race (81.2%) patients than in white patients (77.3%) ( Figure 1A). Among patients with available AFP testing records, black patients had the highest rate (84.1%) of elevated AFP, whereas white patients had the lowest rate (72.5%) of elevated AFP ( Figure 1B). In the subset cohort of HCC patients diagnosed at an early stage (AJCC stage I), the proportion of patients who underwent AFP testing remained higher in the black (79.5%) and other race (82.2%) groups than in the white patients group (78.0%) ( Figure 1C). Meanwhile, the rates of elevated AFP decreased in all three race groups, but black patients still presented the highest rate (79.6%) among them ( Figure 1D).
The period trend of patients who underwent AFP testing increased from 71.5% in 2004-2006 to 80.1% in 2013-2015 in white patients, and from 72.3% to 81.1% in black patients ( Figure 1E) at the same period in the whole cohort. The proportion of elevated AFP decreased from 76.0% to 69.7% in white patients, and from 85.3% to 82.5% in black patients, as determined by comparing the 2004-2006 period to the 2013-2015 period ( Figure 1F). The period trends of the performing rates of AFP testing and elevated AFP rates in the subcohort of early-stage HCC patients were shown in Figure  1G,H, respectively.

| Factors associated with elevated AFP levels according to race
Factors associated with increased AFP levels according to race were analyzed in the cohort with available AFP testing results. Multivariate analysis ( However, the association between age and elevated AFP level was not significant in the black and other race groups. Multivariate analysis showed that higher AJCC stage, lower tumor grade, and higher liver fibrosis score were associated with a higher incidence of increased AFP levels in these three race groups, although the ORs were different among them. Detailed results were shown in Table 2.

| Elevated AFP as a prognostic factor of OS stratified by race
The median OS for the entire cohort with AFP records and survival information was 11 months; the 1-, 3-, and 5-year survival rates were 66.5%, 42.0%, and 31.1%, respectively. AFP-positive patients had a lower median OS (9 months) than those with normal AFP levels (27 months). After stratification by race, the median OS was worse in the AFP-elevated patients than in the AFP nonelevated patients in each race groups, with the values of 7 vs 22 months in the black race group, 9 vs 25 months in the white race group, and 12 vs 39 months in the other race group, respectively ( Figure 2).

| DISCUSSION
Serum AFP level is not considered as the best indicator for HCC according to some guidelines. [13][14][15] This is due to the low sensitivity and high false-negative rate of serum AFP for the diagnosis of HCC, as reported in recent studies. 1,16 In addition, AFP testing is suboptimal because of its cost in routine surveillance of early-stage HCC. 17 However, the Liver Cancer Study Group of Japan 15  To the best of our knowledge, this is the first populationbased study assessing racial disparities among patients undergoing AFP testing and the rates of elevated AFP among patients with HCC in the United States. The present study analyzed a large cohort of patients diagnosed with HCC, and showed significant disparities in the rate of AFP testing performed and elevated AFP rate among three race groups. The results showed that black (79.7%) and other race (81.2%) HCC patients were more likely to undergo AFP testing than white race patients (77.3%). Moreover, significant differences in the incidence of elevated AFP were analyzed according to race, with a significantly higher rate of elevated AFP in black patients than in white patients. A previous study of racial disparities in AFP testing showed that blacks accounted for approximately half (43%) of the patients with HCC and a normal AFP level. 11 Our findings were consistent with the results of the HALT-C trial, which showed that elevated AFP levels were more frequent in blacks. 12 The differences in the results could be due to the variations in methodology, such as patient selection (the previous studies included HCC patients with hepatitis C and both were multicenter based clinical studies), or they could represent differences among patients at different stages of HCC. To clarify this issue, we analyzed the disparities in AFP testing and elevated AFP levels in early-stage HCC patients. After controlling for the stage of HCC, the AFP testing performing rate in black patients was still higher than that of white patients. The analysis of the early-stage HCC cohort revealed a decrease in the proportion of elevated AFP in each race group, however, black race patients remained the group with the highest rate of undergoing AFP testing and proportion of elevated AFP. Our results also indicated that regardless of racial differences, AFP testing performing rates increased, and the positive AFP rates declined in each HCC patients race group during the 2004-2015 period.
Despite the present findings that the black HCC patient group was more likely to have a higher proportion of patients with elevated AFP levels, other disparities associated with elevated AFP levels were observed among the three race F I G U R E 2 Kaplan-Meier curves for OS of HCC patients by AFP level. A, OS in all patients with unelevated or elevated AFP level. B, OS in white race patients with unelevated or elevated AFP level. C, OS in black race patients with unelevated or elevated AFP level. D, OS in other race patients with unelevated or elevated AFP level groups. Female white HCC patients aged 45-64 years, with late AJCC stage, and with larger tumor size were more likely to have elevated AFP levels. However, gender was not a factor associated with elevated AFP levels, and age was not significantly associated with elevated AFP levels in black HCC patients. Unlike the white or black race group, the older patients (≥65 years) in other race group were less likely to have elevated AFP levels. Therefore, disparities in the rate of elevated AFP among HCC patients need to be further analyzed after stratification by race. Although the use of AFP testing for the early diagnosis of HCC is controversial, serum AFP is recommended as a prognostic biomarker for HCC outcome in clinical practice. [19][20][21][22] However, the role of AFP levels at diagnosis as an independent risk predictor associated with OS remains unclear. 23,24 Our data indicate that elevated AFP level is a significant prognostic factor for HCC OS in all three race groups. However, after adjusting for AFP level and other factors, age was not an effective predictor of OS in the other race group, whereas it was a predictor in the white race group. Fibrosis score was not a prognostic factor for OS in black and other race groups after controlled by AFP level. Even after controlling for AFP level and other prognostic factors, disparities in other prognostic factors were still existed among female patients in the other race group.

| Limitations
The present study had a strong feature in its generalizability, as the findings were based on population data, but the study had several limitations. First, we did not have data on the actual numerical value of AFP levels. Normal and elevated AFP values, and a uniform cutoff value were lacking, and we had to rely on local interpretation. However, we believe that similar and standardized reference values are used in most laboratories in the United States. Second, there may have been testing, reporting, and interpretation differences in AFP values between hospitals both within and across the SEER cancer registries. It is not clear to what extent the bias introduced by nonrandom missing data and/or variation in the reporting of AFP status would change our results. Additionally, the potential for mis-classification of race needs to be considered, as the data were extracted from medical records. However, taken together with the comparisons of case proportions and survival by race, our results are robust and informative. The present study provides evidence of racial differences in the use of AFP testing as a biomarker for HCC surveillance, diagnosis, as well as a prognostic factor.

| CONCLUSION
In conclusion, the present study demonstrated that AFP testing is important for the diagnosis and prognosis of HCC in different racial groups. However, racial disparities in the factors associated with elevated AFP levels among the three race groups were existed. We strongly recommend the combination of AFP testing with ultrasonography for the detection of HCC, especially for black race patients. In addition, combining patient demographic characteristics, especially race, with AFP testing would increase the value of AFP testing for HCC surveillance and diagnosis. AFP level remained a prognostic factor for HCC even after accounting for demographics and tumor characteristics; however, further studies are needed to improve the prediction of HCC patient outcomes, especially among racial minorities.