Background: Proteasomes are the main non-lysosomal proteolytic structures which regulate crucial cellular processes. Circulating proteasome levels can be measured using an ELISA test and can be considered as a tumour marker in several types of malignancy. Given that there is no sensitive marker of hepatocellular carcinoma (HCC) in patients with cirrhosis, we measured plasma proteasome levels in 83 patients with cirrhosis (33 without HCC, 50 with HCC) and 40 controls.
Methods and results: Patients with HCC were sub-classified into three groups according to tumour mass. α-Fetoprotein (AFP) was also measured. Plasma proteasome levels were significantly higher in patients with HCC compared to controls (4841 (SEM 613) ng/ml vs 2534 (SEM 187) ng/ml; p<0.001) and compared to patients with cirrhosis without HCC (2077 (SEM 112) ng/ml; p<0.001). This difference remained significant when the subgroup of patients with low tumour mass (proteasome level 3970 (SEM 310) ng/ml, p<0.001) was compared to controls and patients with cirrhosis without HCC. Plasma proteasome levels were independent of the cause of cirrhosis and were weakly correlated with AFP levels. With a cut-off of 2900 ng/ml, diagnostic specificity for HCC was 97% with a sensitivity of 72%, better than results obtained with AFP. Diagnostic relevance of plasma proteasome measurement was also effective in low tumour mass patients (sensitivity 76.2% vs 57.1% for AFP).
Conclusion: The plasma proteasome level is a reliable marker of malignant transformation in patients with cirrhosis, even when there is a low tumour mass.
Statistics from Altmetric.com
Hepatocellular carcinoma (HCC) is the most common malignant tumour of the liver worldwide, with an estimated incidence of one million cases per year.1 It is currently associated with a very poor prognosis. In Western countries, HCC generally develops within the context of chronic liver disease, namely cirrhosis (90% of cases), regardless of aetiology (hepatitis B and C virus, alcohol consumption, auto-immunity or haemochromatosis). In Western countries, the incidence of HCC has increased greatly over the last decade, mainly due to hepatitis C viral infection.
HCC prognosis is strongly correlated with diagnostic delay. Current diagnosis of HCC relies on imaging of the liver tumour2 3 and the detection of α-fetoprotein (AFP) in the patient’s serum.4 All these diagnostic methods are efficient especially for relatively large tumours. Characterisation of novel markers allowing early and accurate diagnosis of HCC appears therefore to be a crucial issue for patient survival, with a clear relevance for public health.
During the past few years, there has been a growing interest in the study of proteasomes in malignant diseases, and proteasome inhibitors now represent a promising new class of anticancer agent in different types of tumour, including HCC.5 6 The proteasome is the main non-lysosomal proteolytic structure of the cell, degrading proteins tagged by the ubiquitin conjugation system. It plays a crucial regulatory function in major cell processes, such as cell cycle regulation, apoptosis, differentiation, DNA repair and degradation of many important rate-limiting enzymes involved in metabolic pathways.7–9 In addition, proteasomes play a key role in the production of antigenic peptides with high affinity to major histocompatibility complex class-I (MHC-I).10 11 Given the wide range of proteins directed towards proteasome degradation, including proteins controlling the cell cycle, it has been demonstrated, as expected, that a high proteasome cell level is observed in early stages of cell differentiation and neoplastic processes.12–14
Proteasomal structures can be detected and their concentration measured in the plasma of normal individuals as well as in patients. Specific enzyme-linked immunosorbent assay (ELISA) tests have been designed for this purpose.15–17 Previous reports have demonstrated the presence of elevated plasma proteasome levels in patients with myeloid haemopoietic malignancies15–17 and solid tumours.16 18
The aim of the present study was to investigate the usefulness of the elevation of plasma proteasome levels in patients with HCC and its potential relevance for early diagnosis of HCC in patients with cirrhosis.
PATIENTS AND METHODS
Eighty-three patients from the Montpellier and Nìmes hospitals were prospectively tested after each had given informed consent. All patients included had cirrhosis previously proven on liver biopsy. The severity of cirrhosis was evaluated using the Child–Pugh classification.19 Patients were then separated into two groups, those without (n = 33) and those with (n = 50) HCC, according to criteria summarised in table 1.20 When histological examination of tumour was performed, grading was assessed according to the WHO criteria.21
In all patients with HCC, tumour diagnosis was recent and no specific treatment had been initiated. Patients with other simultaneous or previous history of cancer were excluded from the study, as well as patients with concomitant inflammatory disease, in order to avoid any interference in interpretation. Three subgroups were defined amongst patients with HCC, partially according to the Milan criteria for liver transplantation in patients with HCC and cirrhosis22 on the basis of the number and/or size of the liver nodules: subgroup 1, only one nodule under 3 cm; subgroup 2, presence of a single tumour between 3 and 5 cm in diameter or no more than three tumour nodules each 3 cm or less in diameter; subgroup 3, tumour mass larger than previous subgroups 1 or 2. There were 21 patients in subgroup 1; 14 in subgroup 2; and 15 in subgroup 3. Forty healthy volunteers served as controls.
Measurement of plasma proteasome concentration
Blood was harvested on citrate anticoagulant, and plasma stored at −80°C after adjunction of glycerol (50% final volume). Plasma proteasome concentration was measured using a standard ELISA technique as previously described.15 16 Briefly, plasma (diluted 1/10) was incubated for 1 h on plates coated with mouse anti-proteasome (alpha C2 subunit) monoclonal antibody (MCP20 clone, diluted 1/4500; Affiniti Research Products, Exeter, UK). Proteasome-captured subunits were detected by a rabbit anti-proteasome polyclonal antibody (I571, diluted 1/1500; kindly provided by KB Hendil, Copenhagen, Denmark), then revealed by goat anti-rabbit peroxidase-conjugated antibody and o-phenylenediamine dihydrochleride (OPD) substrate (Sigma Laboratories, St-Quentin-Fallavier, France). Plates were read at a wavelength of 492 nm, and results were expressed as concentration of proteasome protein in ng/ml after comparison to a standard curve established with purified human 20S proteasome (Affiniti Research Products).
Other biological parameters tested
In patients with cirrhosis, with and without HCC, standard biological parameters were determined: alanine aminotransferase and aspartate transaminase levels, bilirubin and serum albumin levels, prothrombin time, and AFP. AFP was not tested in controls. The upper reference level of AFP in our laboratory was 15 ng/ml.
Biological parameters, namely AFP and proteasome plasma levels, were expressed as mean with the standard error of the mean (SEM). The χ2 test or Yate’s test or Fisher’s exact test were used for comparison of qualitative variables; and the unpaired Student t test or Wilcoxon rank sum test was used for comparison of quantitative variables between groups, according to data distribution. Correlations were studied using the Spearman rank order test.
The outcome variable was the presence of HCC. The following diagnostic indices were calculated for AFP and proteasome plasma levels: accuracy (or efficiency or matched), sensitivity, specificity, positive predictive value, negative predictive value, and Youden’s index, J (where J = sensitivity + specificity − 1).
Receiver operating characteristics (ROC) curves and the respective areas under the curves (AUCs) were calculated. The significance level was p<0.05. Logistic regression analysis was used to investigate the relationship between the level of plasma proteasome and various explanatory variables. Optimal cut-off was determined by ROC curves, which corresponded to the lowest number of mismatched (or misclassified) samples (sum of false positives and false negatives).
An internal cross-validation of diagnostic studies was achieved by the bootstrap method.23
Eighty-three patients with cirrhosis were prospectively enrolled in the study, 33 without and 50 with HCC. The main clinical and biological parameters can be seen in table 2. The control group consisted of 19 men and 21 women, age 56.1 (SEM 2.1) years. There was no statistically significant difference between controls and patients with cirrhosis without HCC as regarding age (p = 0.891). In contrast, the difference was statistically significant both between controls and patients with HCC, and between patients with cirrhosis with and without HCC (0.010 and 0.014, respectively).
Plasma proteasome levels did not depend on the aetiology of cirrhosis
When comparing plasma proteasome levels, no statistical difference could be seen according to the cause of cirrhosis, in the entire population of 83 patients, as well as in the different groups (cirrhosis, all HCC cases, HCC subgroups). Likewise, no statistical difference was found for plasma proteasome levels according to the severity of cirrhosis as evaluated by Child–Pugh scoring.19
Plasma proteasome levels in patients with HCC were significantly higher than in controls and in patients with cirrhosis without HCC (fig 1 and table 3)
The mean plasma proteasome level observed in the control population was 2534 (SEM 187) ng/ml (range, 654–6446 ng/ml). In the HCC patient population, the mean proteasome level was 4841 (SEM 613) ng/ml (range, 736–23879 ng/ml), which was significantly different from controls (p<0.001). In patients with cirrhosis without HCC, the mean plasma proteasome level was 2077 (SEM 112) ng/ml (range, 600–3443 ng/ml), which was lower but not statistically different to that observed with controls (p = 0.101).
When comparing patients with and without HCC, the difference in plasma proteasome level remained highly significant (p<0.001).
When considering specifically patients with HCC with low tumour mass (subgroup 1), the mean plasma proteasome level was 3970 (SEM 310) ng/ml; range, 2301–7374 ng/ml. This result was significantly different from what was seen in the control population and in patients with cirrhosis without HCC (p<0.001 for both).
Plasma proteasome levels were not significantly different in the three HCC subgroups defined according to tumour mass (fig 1 and table 3)
As compared to subgroup 1 patients, subgroup 2 patients (4277 (SEM 824) ng/ml; range, 736–13 734 ng/ml) and subgroup 3 patients (6588 (SEM 1815) ng/ml; range, 1305–23 879 ng/ml) did not exhibit a significant difference, although mean values were increasing from subgroups 1 to 3. There was a statistical difference in plasma proteasome levels between each HCC subgroup with both controls (p<0.001, p = 0.011 and p = 0.0025, respectively) and patients with cirrhosis without HCC (p<0.001).
Plasma proteasome levels did not depend on histological grading of HCC
Plasma proteasome levels were 3820 (SEM 1114) ng/ml for undifferentiated tumours and 4965 (SEM 4743) ng/ml for differentiated tumours (p = 0.370).
Correlation between proteasome and AFP levels
The values of AFP detected in the cirrhotic population without HCC (mean, 8.4 (SEM 2.3) ng/ml; range, 1.2–80.2 ng/ml) were below the limit of normality in our laboratory (15 ng/ml) except for two patients (15.1 and 80.2 ng/ml, respectively). When considering the 50 patients with HCC, the mean AFP level was clearly above the normal value (2062 (SEM 1367) ng/ml; range, 1.9–58 267 ng/ml), and statistically higher than in patients with cirrhosis without HCC (p<0.001). When considering HCC subgroups, AFP mean level was clearly correlated with tumour mass. It was not significantly increased in the subgroup 1 (mean, 24.3 (SEM 14.8) ng/ml; range, 1.9–316.8 ng/ml; p = 0.269) but was significantly higher in subgroup 2 (mean, 2898 (SEM 2650) ng/ml; range, 3.5–37 300 ng/ml, p<0.001) and in subgroup 3 (mean, 4135 (SEM 3868) ng/ml; range, 3.2–58 267 ng/ml; p<0.001) as compared to patients with cirrhosis but without HCC. Patients in subgroups 2 and 3 exhibited a significantly higher AFP level than subgroup 1 patients (p = 0.018 and p = 0.002, respectively).
There was a positive correlation between plasma proteasome levels and AFP levels in the group of 83 patients with cirrhosis (p = 0.03) but with a poor coefficient of correlation (rS = 0.279; data not shown).
Diagnostic relevance of plasma proteasome measurement (table 4)
In order to determine the optimal cut-off value minimising the number of misclassified samples, we first established the ROC curve (AUC = 0.875). In our study, this cut-off value was 2900 ng/ml. In the global patient population, 68/83 patients (81.9%) were correctly classified as regarding the presence of HCC. In the group of patients with cirrhosis without HCC, 32/33 had a proteasome level below this cut-off, providing a specificity of 97%. There was only one case in this group with a proteasome level above 2900 ng/ml. In the HCC population, 36/50 patients with HCC were above the cut-off value, corresponding to a sensitivity of 72% with 14 misclassified cases (28%). In summary, only one patient with a plasma proteasome level above 2900 ng/ml did not have HCC. Among the 37 patients predicted to have HCC by the test, 36 did indeed have HCC, which corresponds to a very high positive predictive value (PPV; 97.3%). Among the 46 patients below the proteasome cut-off level and thus predicted negative for HCC, 32 were true negatives, corresponding to a negative predictive value NPV of 69.6%. These values of PPV and NPV exceed, respectively, positive and negative prevalence (60.2% and 39.8%, respectively) by 37.1% and 29.8%, proving they were highly relevant.
Plasma proteasome levels provide better diagnostic indices than AFP measurement (table 4)
The ROC curve (AUC = 0.745) for AFP for the entire patient group determined a cut-off value of 4.65 ng/ml. Based on this cut-off, 57 patients were correctly classified regarding the presence of HCC (68.7%), a lower proportion when compared to proteasome. In the group of patients with cirrhosis without HCC, 18/33 had a low AFP value, which corresponds to a specificity of 54.5%. In the HCC group, 11 patients with HCC presented a low AFP level, which corresponded to a sensitivity of 78%. The positive predictive value was 72.2%, exceeding positive prevalence by only 12.0%, so this index was not relevant. Negative predictive value was 62.1%, 22.3% up to negative prevalence, meaning it was borderline relevant.
Effectiveness of combining measurements of proteasome and AFP levels (table 4)
Sixty-two patients exhibited at least one of the two tested values above the respective cut-off limits. This population contained 16 false positive patients (patients with cirrhosis without HCC). Only four patients with HCC (8%) had both values below the respective cut-off. The use of proteasome and AFP measurements for the diagnosis of HCC therefore provided a specificity of 51.5% and a sensitivity of 92%, and thus a Youden’s index largely lower than with proteasome alone. Because of the numerous false positives, the PPV was not relevant (74.2%), only 14.0% up to the positive prevalence. NPV was highly relevant (81.0%), with 41.2% up to negative prevalence.
In summary, this combination was not effective, although the sensitivity was improved, yet the other major indices were lowered by AFP contribution.
Diagnostic relevance of plasma proteasome measurement was also effective in low tumour mass patients (table 5)
When considering the sole HCC group with low tumour mass, proteasome measurement continued to provide highly relevant diagnostic indices. Accuracy increased up to 88.9%, sensitivity increased to 76.2%, thus the Youden’s index increased to 0.732. PPV and NPV were highly relevant, respectively 94.1% and 86.5%, resulting in a 55.2% and 25.4% increase compared to positive and negative prevalence (38.9% and 61.1%, respectively).
In contrast, AFP measurement gave poor results in this group, sensitivity decreased to 57.1% (only 12/21 patients were above the normal limit), the Youden’s index became very low (0.117). PPV was weak (44.4%) and not relevant (+5.5% than positive prevalence). NPV was slightly higher than in the total HCC population (66.7%) but only 5.6% more than negative prevalence, and was therefore irrelevant.
In low tumour mass also, the combination of AFP and proteasome parameters was not accurate because of the poor contribution of AFP.
Internal cross-validation of diagnostic cut-offs (table 6)
An internal cross-validation of diagnostic studies was achieved by the bootstrap method. The bootstrap is a statistical inferential technique based on successive re-samplings. By using the optimal cut-offs for proteasome and AFP computed on each of the 1000 bootstrapped samples, 1000 values for percentage of matched predictions, sensitivity and specificity were obtained for proteasome alone, AFP alone and for the combination of both measurements. The median value and a 95% variability interval were then computed and compared to values obtained on the original sample.
The optimal proteasome cut-off value and the corresponding percentage of matched values were stable by re-sampling. Cut-off and percentage of matched values for AFP were underestimated in comparison to the median obtained on the 1000 bootstrapped samples. Consequently, this underestimation was found when combining proteasome and AFP.
In the present study, we show that plasma proteasome levels were significantly increased in patients with HCC, when compared to controls and patients with cirrhosis without HCC. Patients with cirrhosis with HCC were older than patients with cirrhosis without HCC and controls but the difference is quite logical since HCC is a long-term complication of cirrhosis. Moreover, we showed in previous studies that plasma proteasome levels are not correlated with age.16 Thus, plasma proteasome level seems to be a reliable tumour marker for HCC in patients with cirrhosis.
We and others have shown the potential interest of circulating proteasome as a tumour marker in haemopoietic malignancies, namely myeloid disorders, myeloma and acute leukaemias as well as in solid tumours.15–18 24 25 Additionally, circulating proteasome levels are also increased in HCC. From a diagnostic point of view, one has to remain aware that our patient group excluded patients with concomitant cancers of other origin, where an increased proteasome level is no longer specific for HCC. Malignant diseases are not the only conditions in which plasma preoteasome can be increased (for a review see Sixt and Dahlman24). All this is to be taken into account for routine diagnostic purposes. It is clear that the diagnostic indices of plasma proteasome measurement for HCC provided here would only be valid for a population with cirrhosis.
Measurement of plasma proteasome level seems therefore to be a relevant and original tool for detection of HCC in patients with cirrhosis. This novel marker appears to provide a high PPN (97.3%) and NPV (69.6%), better than corresponding values obtained with AFP.
It is noticeable that, in HCC, the increase in plasma proteasome level could be detected even in patients with a low tumour mass, in contrast to what has been observed for other tumours.16 18 It can be remarked, however, that the mean level of circulating proteasome increased from subgroup 1 to subgroup 3. The lack of a statistically significant difference between the three subgroups could be related to the sample size.
In contrast, AFP measurement allowed detection of less than half the cases of HCC (48%), this result being consistent with previous reports.26 AFP levels were clearly related to tumour mass. It is noteworthy that, in our population, patients with a low tumour mass were relatively over-represented, which explains why the cut-off value for AFP (4.65 ng/ml) is clearly lower than the usual normal value (<15 ng/ml), which is a disadvantage for proteasome in the comparison with AFP as a tumoural marker. Nevertheless, proteasome retained its comparative advantage. The combined use of both parameters did not allow an improvement in the diagnostic indices.
Recent data concerning the molecular biology of HCC has contributed to understanding the underlying mechanism of proteasome increase in the plasma of cancer patients. Neo and colleagues27 have demonstrated using a minimal dataset approach of gene profiling an increased expression of several genes of the ubiquitin/proteasome pathway in HCC tumour samples as compared to normal liver tissue. Out of 218 significantly modified genes in HCC tumour samples, the most remarkable were the 20S alpha 6 subunit (PSMA1) and 20S beta 7 subunit (PSMB4) proteasome genes.27 It is noticeable that in our ELISA test the capture is performed using an anti-HC2 monoclonal antibody (directed against the PSMA1 subunit). More recently, Cui and colleagues28 have shown in HBx gene knock-in transgenic mice, which develop HCC at the age of 18 months, an upregulation of several proteasome subunits when compared to wild-type control mice: 20S alpha 1 (PSMA6), 20S beta 7 (PSMB4), 19S ATPase Rpt1 (PSMC2) and 19S non-ATPase Rpn5 (PSMD12). It can therefore be hypothesised that overexpression of the proteasome subunit could be a crucial step of malignant transformation in hepatocytes. It can be remarked that plasma proteasome levels do not depend on the type of underlying cirrhosis. The increase observed seems therefore related to common steps in the malignant process.
Proteasome inhibitors are a new class of anticancer agents, which are now widely used in several tumour types, namely multiple myeloma.29 They have been tested in HCC models and shown to be efficient, for instance, in conjunction with TNF-related apoptosis-inducing ligand (TRAIL).6 This therapeutic relevance reinforces the hypothesis that proteasome plays a crucial role in HCC onset and development. Plasma proteasome evaluation could thus become a highly relevant tool for treatment monitoring in the future. Taken together, the results provided by the present study demonstrate that plasma proteasome is a novel and relevant marker of HCC, namely for early detection of HCC in patients with cirrhosis.
Competing interests: None.
Funding: This work was supported by grants from the Ligue Nationale contre le Cancer (comités du Gard et des Pyrénées Orientales).
Ethics approval: Approval for this study was obtained from the Institutional Review Board of Nìmes Hospital on 25 June 2008.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.