The incidence of primary liver cancer (hepatocellular carcinoma, HCC) has been gradually increasing in recent years.HCC has a poor prognosis, with a 5-year survival rate of less than 5%, and approximately 59.18 million people die from HCC each year worldwide. HCC is the second most common cancer killer in China, accounting for more than half of all HCC deaths in the world each year. One of the reasons for poor prognosis of hepatocellular carcinoma is that most patients are not detected and diagnosed at an early stage. Therefore, early detection, early diagnosis and early treatment are important factors to improve the survival and prognosis of HCC patients. Alpha-fetoprotein (AFP) is currently the only widely used serum marker for HCC diagnosis, and AFP is used for the screening of high-risk groups of hepatocellular carcinoma, which has significantly improved the postoperative survival and prognosis of some patients. In this study, we investigated the application and value of AFP in the clinical diagnosis of hepatocellular carcinoma by analyzing the relationship between serum AFP levels and clinical characteristics of hepatocellular carcinoma patients in different populations, so as to have a better understanding of the distribution and characteristics of AFP in HCC patients and further improve the early diagnosis rate and diagnostic accuracy of HCC. Subjects and methods 290 patients with HCC admitted to our hospital from August 2002 to November 2003, including 254 males (87.6%) and 36 females (12.4%), aged 13-83 years, with an average age of 50.17 ± 11.01 years, were selected. Forty-eight patients with cirrhosis were admitted to the gastroenterology department of our hospital, including 34 (70.8%) males and 14 (29.2%) females; age ranged from 26 to 76 years, mean (52.23 ± 10.26) years. Forty-nine patients were examined in our physical examination center, of whom 33 (67.3%) were male and 16 (32.7%) were female; age ranged from 21 to 62 years, with an average of 38.71 ± 10.23 years. Methods Serum AFP levels were analyzed in HCC patients, cirrhotic patients and healthy check-ups; the diagnostic sensitivity and specificity of AFP in HCC and cirrhotic patients were analyzed at different thresholds; serum AFP levels were analyzed in HCC patients with different tumor sizes, different tumor numbers and different clinical stages of TNM of the International Union Against Cancer. Statistical analysis was performed using Graphpad p rism 4 software, including U-test and ANOVA, etc. P < 0.05 indicated significant differences. The median (25th and 75th percentiles) serum AFP level in HCC patients was 158.2 ( 10.5, 1597) μg/L, with a distribution range of 1 to 60500 μg/L. The median (25th and 75th percentiles) serum AFP level in cirrhotic patients was 45.24 ( 19.4, 237.3) μg/L, with a distribution range of 2.2 to 2 010 μg/L. The median (25th and 75th percentiles) serum AFP levels in healthy subjects were 4.9 ( 1.5, 8.7) μg/L, with a range of 0 to 30.2 μg/L. Blood levels between patients with HCC and patients with cirrhosis (P = 0.0274), between patients with HCC and healthy subjects (P = 0.0001), between patients with cirrhosis and healthy subjects (P = 0.0001), and between patients with cirrhosis and healthy subjects (P = 0.0001), were higher than those of patients with cirrhosis. The differences in serum AFP levels between patients with HCC and patients with cirrhosis (P = 0.0001), and patients with cirrhosis and healthy subjects (P = 0.0001) were all significant. The diagnostic sensitivity and specificity of AFP in HCC and cirrhotic patients at different diagnostic thresholds When the diagnostic threshold of AFP was set at 20 μg/L, 195 (67.2%) of HCC patients were diagnostically predicted to be positive and 95 (32.8%) were diagnostically predicted to be negative; 14 (29.2%) of cirrhotic patients were diagnostically predicted to be negative and 34 (70.8%) were diagnostically predicted to be positive. The diagnostic sensitivity was 67.2 %. The diagnostic sensitivity was 67.2%, the specificity was 29.2%, the accuracy of positive prediction was 85.2%, and the accuracy of negative prediction was 12.8%. When the diagnostic threshold of AFP was set at 400 μg/L, 124 (42.8%) of HCC patients were diagnosed as positive and 166 (57.2%) were diagnosed as negative; 43 (89.6%) of cirrhotic patients were diagnosed as negative and 5 (10.4%) were diagnosed as positive. The diagnostic sensitivity was 42.8%, the specificity was 89.6%, the accuracy of positive prediction was 96.1%, and the accuracy of negative prediction was 20.6%. Serum AFP levels in HCC patients with different tumor sizes HCC patients were divided into two groups according to tumor size: tumor ≤5 cm and tumor >5 cm. The median (25th and 75th percentiles) serum AFP levels in patients with tumor ≤5 cm (177 cases, 61%) were 98.1 (11.75, 1019) μg/L, with a distribution range of 1.04-25530 μg/L. The median (25th and 75th percentiles) serum AFP level was 589.4 ( 9.215, 3568) μg/L in patients with tumors > 5 cm (113 patients, 39%), with a distribution range of 1 to 60500 μg/L. The difference in serum AFP levels between the two groups was significant (P = 0.0009). Serum AFP levels in HCC patients with different number of tumors were divided into two groups according to the number of tumors, The median (25th and 75th percentiles) serum AFP level was 1120 ( 46.04, 2063 ) μg/L, with a distribution range of 2.35-60500 μg/L. The difference in serum AFP levels among patients with HCC with different tumor numbers was significant (P = 0.0001). Serum AFP levels in HCC patients with different TNM stages HCC patients were divided into three groups according to TNM stages, The difference in serum AFP levels between TNM stage I and II HCC patients was not significant (P = 0.1033), while the difference in serum AFP levels between TNM stage I and TNM stage III-IV (P = 0.0001) and TNM stage II and TNM stage III-IV (P = 0.0003) HCC patients was significant. Discussion Serum AFP levels were significantly higher in patients with HCC than in patients with cirrhosis and in healthy subjects, and in turn were significantly higher in patients with cirrhosis than in healthy subjects. Therefore, elevated AFP levels are of great value for the diagnosis of HCC. When patients with elevated AFP are encountered, attention should be paid to the differentiation of HCC from cirrhosis. Dynamic observation of serum AFP levels in patients with chronic hepatitis and cirrhosis showed that the increase in serum AFP levels is usually transient or fluctuates repeatedly, while the increase in serum AFP in patients with HCC is usually stable and persistent, which can be differentiated by combining with imaging. In addition, elevated serum AFP levels can also be seen in pregnancy, chronic active hepatitis, germline tumors, teratoma, other gastrointestinal tumors, and metastatic liver cancer, and should be distinguished. The clinical application of AFP as a tumor marker for HCC often encounters the problem of setting the diagnostic threshold. When using 20 μg/L as the diagnostic threshold, the diagnostic sensitivity is high (67.2%), but the specificity is low (29.2%). In contrast, when 400 μg/L was used as the diagnostic threshold, the diagnostic prediction specificity improved significantly (89.6%), but the sensitivity decreased significantly (42.8%). In order to improve the diagnostic sensitivity, 20 μg/L is now used as the diagnostic threshold in clinical application, and it is used together with other imaging tests for screening of high-risk groups, which has achieved good results. However, even if 20 μg/L is used as the diagnostic threshold, 30%-40% of HCC patients are still AFP negative (AFP < 20 μg/L). In addition to being a clinical diagnostic marker, AFP can be used as an indicator for postoperative recurrence and metastasis monitoring. For AFP-positive hepatocellular carcinoma treated with radical surgery, AFP should be reduced to negative within a certain period of time. In case of postoperative recurrence or metastasis, AFP may rise again. However, it should be noted that a certain percentage of patients with original AFP-positive HCC can also be AFP-negative when they recur, while patients with AFP-negative HCC can be AFP-positive when they recur. Therefore, postoperative monitoring should be combined with imaging. The analysis of serum AFP levels in HCC patients with different tumor sizes showed that the level of AFP correlated with tumor size, and the level of AFP could reflect the size of tumor to some extent. The serum AFP levels of HCC patients with multiple tumor nodules were higher than those of HCC patients with single tumor nodules. The analysis of serum AFP levels in HCC patients with different TNM stages showed that the serum AFP levels in patients with TNM stages III-IV were higher than those in patients with stages I and II. This suggests that AFP levels can reflect the early and late stages of the disease. However, the AFP level of small liver cancer and early liver cancer is low, so the sensitivity of AFP in the diagnosis of small liver cancer and early liver cancer is low, and it is not a good early diagnostic index. It is necessary to study new diagnostic markers for liver cancer and apply them together with AFP to improve the diagnostic accuracy and early diagnosis rate of HCC, especially AFP-negative liver cancer, small liver cancer and early liver cancer, so as to improve the prognosis of HCC. In conclusion, AFP has an important value in the clinical diagnosis of HCC, but a comprehensive and objective understanding and evaluation of it is an important prerequisite for its clinical diagnostic value.