Hepatocellular carcinoma (HCC) is one of the most frequent types of cancer worldwide, with an incidence of 150 cases per 100 000 inhabitants per year in Asia and Africa and ∼5 cases per 100 000 per year in Europe, North America, and Australia (1). HCC can occur in healthy livers, but it is generally associated with hepatic disease. Cirrhosis is the most important risk factor for its development (2). Many studies have been dedicated to the early detection of HCC in patients with chronic liver disease (CLD), based on semiannual or annual measurement of serum α-fetoprotein (AFP) and ultrasound study of the liver. The interpretation of AFP results usually involves comparison with a fixed cutoff ranging from 20 to 500 μg/L, with sensitivities ranging from 33% to 64% (3)(4)(5). Another approach would be to assess changes in consecutive results from the same patient. To establish a criterion for dynamic assessment of a specific biological constituent, the first step is to define when a difference between two consecutive results indicates a change in a patient’s health status. The most widely accepted approach for this purpose is the so-called reference change value (RCV), a concept described by Harris and Brown (6). Using serial analytic results from the same individual for a specific constituent, it is possible to calculate the RCV that defines how large a difference between two consecutive determinations is statistically significant (P <0.05). The RCV encompasses both biological and analytical variation. The RCV has been determined for various constituents in healthy persons and in some pathologies (7)(8)(9)(10)(11)(12)(13). In diseases that can modify the homeostatic set point and the degree of fluctuation of a specific constituent, the within-subject biological variation (BV) may be different from that of persons who do not have these diseases (13)(14)(15). This is seen in AFP determinations for the early detection of HCC in patients with CLD. These patients present with higher AFP values than healthy individuals and with abnormal patterns of AFP release because of processes of hepatic necrosis/regeneration (16)(17). The index of individuality (I.I.) of an analyte indicates whether significant changes in a patient’s health status can be detected by comparing the analytic results with a population-based cutoff value or whether each patient should be evaluated against him- or herself (18). The aims of this study were as follows: (a) to estimate the BV of AFP in patients with hepatic disease and in patients with nonhepatic conditions, comparing the two groups; (b) to establish analytic quality goals for AFP determinations; and (c) to calculate the RCV of AFP and evaluate its application in the early detection of HCC in a separate group of CLD patients. Two subgroups, a non-hepatic and a hepatic disease group, were studied to determine the components of analytical variation and BV for AFP. The RCV value obtained was then validated by applying it to a third group consisting of patients with CLD who were included in a program for early detection of HCC. Thirty patients (15 men and 15 women) with colon adenocarcinoma (Astler–Coller classification stages A or B1) treated surgically were studied to determine the BV components for AFP. For 3 years after the procedure, three samples from each patient were assayed every 6 months. During the study period, there was no evidence of recurrence. BV values for AFP were also determined in 85 clinically stable patients, 40 with chronic hepatitis and no clinical evidence of cirrhosis and 45 with evidence of cirrhosis, classified according to the Child–Pugh Index (CPI) as A (18 cases), B (15 cases), or C (12 cases). The RCV value obtained for AFP was applied to 124 CLD patients [67 men and 57 women; mean (SD) age, 60.4 (13.7) years] from July 1997 to June 2001. The CLD etiologies included 67 with hepatitis C viral infections, 48 with alcoholism, and 9 with miscellaneous conditions; 49 patients had chronic active hepatitis and no clinical evidence of cirrhosis, and 75 had clinical evidence of cirrhosis (CPI A, n = 35; CPI B, n = 26; CPI C, n = 14). The clinical criteria to define cirrhosis included CLD and one or more of the following signs: ascites, encephalopathy, upper digestive tract bleeding attributable to esophageal varices, portal hypertension as determined by ultrasound or endoscopy, bilirubin >25 μmol/L, albumin <36 g/L, and prothrombin time (international normalized ratio) >1.3. The sensitivity and specificity of the RCV were compared with two cutoff values for AFP (20 μg/L, for high sensitivity; and 200 μg/L, for high specificity) that are widely reported in the literature (5)(19) The mean monitoring period was 20 months (range, 12–48 months), with AFP measurements and abdominal ultrasound studies every 6 months. The mean number of determinations per patient was 4.4 (range, 3–10). When the difference between two consecutive AFP results was higher than the RCV criteria but ultrasound findings were normal, abdominal computed tomography (CT) was performed. HCC was diagnosed by ultrasound- or CT-guided fine-needle puncture aspiration of the lesion and histologic study. In addition, one patient suspicious by ultrasound, CT scan, and AFP >200 μg/L was entered in the liver transplant program. HCC was diagnosed by observation of the lesion in the extracted organ. Serum AFP concentrations were measured by electrochemiluminescent analysis on an ELECSYS 2010 instrument (Roche Diagnostics). Samples from the non-hepatic group were tested in a single analytical run. Samples from the hepatic disease group were determined on the same day of collection. Statistical analyses were performed with the Kolmogorov–Smirnov and Mann–Whitney U-tests. Within-run imprecision, as the CV, was 2.5% (n = 10) for a mean AFP value of 4.1 μg/L (obtained with replicates of a patient sample). Between-run imprecision was 7.1% (n = 720) for a mean AFP value of 8.1 μg/L obtained with Precinorm (Roche Diagnostics). The variations of AFP values around the mean in the two study groups (non-hepatic and hepatic disease) did not follow a gaussian distribution; thus, logarithmic transformation was applied (9). Patients, samples, AFP values, components of analytical and biological variation, and the calculations derived from these values. All RCVs, including the RCV of the non-hepatic group, were calculated assuming a CVa of 6.1%. Significant difference (P <0.01) compared with patients with liver disease. Within-run CV. Between-run CV. Patients, samples, AFP values, components of analytical and biological variation, and the calculations derived from these values. All RCVs, including the RCV of the non-hepatic group, were calculated assuming a CVa of 6.1%. Significant difference (P <0.01) compared with patients with liver disease. Within-run CV. Between-run CV. Fifteen of the 124 patients in the validation group were diagnosed with HCC. In four patients, the first sign of HCC was a significant increase in AFP concentrations as determined by the RCV, giving a sensitivity of 26.7% as the first sign. A simultaneous increase in AFP and ultrasound features indicative of HCC were present in 5 of the 15 HCC cases detected; 1 patient had values of 5–18 μg/L (<20 μg/L), with a significant RCV. Six of 15 presented with ultrasound findings as the first sign; 5 of these did not show an increase in AFP over the RCV (or >20 μg/L), and 1 presented with an increase (from 24 to 40 μg/L) that was below the RCV. Among the four cases in which an increase in AFP was the first sign of HCC, abdominal CT disclosed tumors smaller than 3 cm in three patients; in the fourth patient, the tumor was seen on the following ultrasound control at 6 months. The sensitivity and specificity of the RCV for detecting HCC compared with those of the corresponding fixed cutoff values in the validation group are shown in Fig. 1A . The diagnostic efficacy was also calculated, based only on the subset of patients with AFP values >20 μg/L (Fig. 1B ). Specificity ( ) and sensitivity ( ) of the RCV and cutoff to predict HCC in all patients in the program for early detection of HCC (A) and in patients with AFP >20 μg/L (B). AFP measurements and ultrasound examination of the liver have been used to detect small, asymptomatic tumors in cirrhotic patients, allowing application of effective therapy with increased survival rates (20)(21). The laboratory can provide early evidence of changes in the health status of these patients if it is able to maintain its between-run imprecision below one-half the within-subject BV (22)(23)(24)(25) for AFP. The incidence of HCC in CLD obtained in this study was 3.02%, similar rates reported for the northern Mediterranean geographic area (3)(21). These additional findings are worthy of mention: Patients with CLD had higher AFP concentrations and CVw values than patients with the other clinical conditions studied. The components of BV for AFP in healthy individuals is unknown (26). The I.I. obtained for all the groups was <0.6, indicating that the best approach for detecting changes with this analyte is to evaluate each patient against him- or herself by means of the RCV. The sensitivity and specificity for detection of HCC obtained with AFP cutoff values (20 and 200 μg/L) was similar to that reported by others (3)(4)(5)(17)(19). Application of the AFP RCV in the validation group detected significant changes that were later confirmed to be HCC. The sensitivity for detecting HCC using the RCV was similar to that obtained with a cutoff of 20 μg/L. Application of the RCV to the subgroup of patients with at least one AFP result >20 μg/L increased sensitivity and specificity. If the RCV from patients without hepatic disease (RCV = 31.2%) had been used, specificity would have been low (74.7%). This indicates that use of the RCV for monitoring patients requires previous establishment of the within-subject BV for the specific pathology under study. In conclusion, the RCV is a useful tool for dynamic monitoring of AFP in patients with hepatic disease, demonstrating good sensitivity and specificity that increase when applied to patients with high AFP values.
No takes yet. Share an insight, caveat, or question.
Trapé et al. (2003) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: