Hepatocellular Carcinoma Surveillance in a European Setting
Hepatocellular Carcinoma Surveillance in a European Setting
In total, 633 patients were registered from January 1st, 2005 to December 31st, 2012. Of these, seven patients were excluded due to incorrect diagnoses and six cases because of treatment received in other hospitals. In four cases exclusion was made due to missing clinical data in the patient files. The remaining 616 patients were included in the study, however, two of which were lost to long-term follow up due to emigration; hence, they were censored at their last clinical contact. Of the 616 patients, 88 patients (14%) had been referred to Karolinska from centres outside Stockholm County ('external referrals'), whereas 528 patients (86%) were residents in Stockholm County. The total number of newly diagnosed HCCs in Stockholm County 2005–2012 was 703 patients; thus, the Karolinska registry covered 75% of all cases diagnosed in the catchment area during this time period.
Patient characteristics are shown in Table 1. Median age at diagnosis was 65.3 years. Of these were 39 (6.3%) <50 years of age, and 52 (8.4%) >80 years of age. One-hundred and forty-seven patients (24%) were female, 468 (76%) were male. Five-hundred and five patients (82%) had cirrhosis, of whom 266 (53%) were Child–Pugh A, 185 (37%) Child–Pugh B, and 44 (8.7%) Child–Pugh C. In 10 cases (2%) Child–Pugh score data were missing. Two-hundred and twenty patients (36%) had diabetes mellitus type 2, and 163 (26%) had the metabolic syndrome. Data for the calculation of BMI was available in 527 (86%) of patients. Median BMI was 26.3. It exceeded 25 in 330 patients (54%) and 30 in 122 patients (20%). Fourteen patients had BMI <19. Median BMI in patients with NAFLD was 30.8, compared to 25.7 in patients with other underlying diagnoses (P < 0.01).
As shown in Table 1, the total number of patients increased over time, from 241 in Period 1 to 375 in Period 2, and the number of external referrals increased from 28 to 60 patients. The catchment area coverage was 75% in both Period 1 and 2. The proportion of patients with BCLC-stage B was significantly higher in Period 2, whereas the other BCLC-stages showed no significant change over time. The proportion of Child–Pugh classes was similar between Period 1 and 2. Tumour size and performance status were unchanged, but the number of tumours was significantly higher in Period 2 compared with 1 (3.4 vs. 3.0).
Median BMI increased significantly between time periods, from 25.7 to 26.7 (P < 0.05). The number of patients with NAFLD doubled from 22 in Period 1 to 45 patients in Period 2, increasing the proportion of NAFLD from 9% to 12% of the total cohort. This trend did not reach statistical significance. No significant change was seen among the other diagnoses between Period 1 and 2.
Surveillance data of the entire cohort and from the two time periods are summarized in Table 1. In short, 134 (22%) of the HCCs were discovered through surveillance. Twenty-eight (4.5%) of these patients were under surveillance according to patient files, but regularity of the investigations could not be controlled as they were referred from other hospitals, and the radiology data were not available. In about half of the surveillance cases, surveillance was incomplete according to our definition.
In 250 patients (41% of the entire cohort), no indication to perform surveillance was found, i.e. there was an absence of either cirrhosis, chronic hepatitis B, or porphyria; or curative therapy for HCC was considered contraindicated, i.e. due to co-existing cardiovascular or pulmonary disease, ongoing abuse or poor general condition. Hence, in these 250 patients the previous decisions not to perform surveillance were considered correct. In 1.1% of cases surveillance was not carried out because of deficient patient adherence. As seen in Table 1, in the remaining 34% of cases, surveillance was indicated but not carried out because of undiagnosed liver disease, undiagnosed cirrhosis, or doctor's failure to order. Of these, 108 patients (18%) had an unknown liver disease, however, in retrospect were considered suitable for surveillance, had their condition been known. Patient records showed that 72 of these 108 patients with unknown liver disease had paid previous visits to a physician under diagnoses such as hypertension, diabetes, alcohol abuse, obesity and elevated liver function tests. Twenty-four patients (3.9%) had a known liver disease (19 of whom had chronic HCV) but undiagnosed cirrhosis. In 79 cases (13%) the cause for not including the patient in a surveillance programme was unclear, and this group was labelled 'Doctor's failure to order'. In 14 of the patients (2.3%) surveillance data was missing. Surveillance patterns were similar in Period 1 and 2.
In three cases the HCCs were missed by ultrasound surveillance, but diagnosed by computerized tomography.
The number of resections and local ablative therapies increased in Period 2 compared with Period 1, whereas the number of liver transplantations was constant. The largest increase was seen in palliative treatments in Period 2 compared with Period 1 (P < 0.05). Treatment with TACE increased from 32 (13%) to 90 (24%) and sorafenib increased from 17 (7%) to 52 (14%). There were no significant differences in overall survival, or in survival of the respective treatment groups, between Period 1 and Period 2 (data nor shown) (Table 1).
As seen in Table 2, patients in whom surveillance was indicated but not carried out because of undiagnosed liver disease or cirrhosis, or doctor's failure to order, had significantly larger tumours, a smaller proportion of tumours within UCSF criteria, and they received curative treatment less often as compared with patients who had undergone surveillance (P < 0.05).
There was a trend of larger tumours and smaller proportion within UCSF criteria in patients who underwent incomplete surveillance compared with those receiving complete surveillance; however, the proportion of patients receiving curative treatments was similar (57% vs. 58%) (data not shown).
Survival was significantly better in patients with HCC diagnosed through surveillance compared with no surveillance (Fig. 1A). In the group of patients in whom surveillance was indicated but not carried out because of undiagnosed liver disease or cirrhosis, or doctor's failure to order, survival was worse compared with those who underwent surveillance (Fig. 1B). There was no difference in survival between those who underwent complete vs. incomplete surveillance (data not shown).
(Enlarge Image)
Figure 1.
A. Cumulative proportion surviving (Kaplan–Meier) stratified according to 'Surveillance' (n = 134) or 'No surveillance' (n = 468) (data from the whole patient cohort) (P < 0.05). B. Cumulative proportion surviving (Kaplan–Meier) stratified according to 'Surveillance' (n = 134) or 'No surveillance' only in patients in whom surveillance was indicated (n = 211) (P < 0.05). Here, 'No surveillance' implicates that there was an indication for surveillance but it was not carried out because of undiagnosed liver disease, undiagnosed cirrhosis, or doctor's failure to order.
The spectrum of underlying liver disease is demonstrated in Table 3. Forty percent of patients had an underlying hepatitis C, and 17% of these also had a concurrent alcoholic liver disease. Alcoholic liver disease alone was found in 18%, NAFLD in 11% and hepatitis B in 7.5%. Approximately, 10% of patients had a previously healthy liver.
The mean age at diagnosis was highest in patients with NAFLD and in those with previously healthy livers. The metabolic syndrome was significantly more common in patients with NAFLD and alcoholic liver disease as compared with the other underlying etiologies (P < 0.05). The proportion of patients with underlying cirrhosis was lowest in those with NAFLD (healthy livers excluded) (P < 0.05). In patients with NAFLD or alcoholic liver disease, the HCCs were discovered by surveillance in 13% of cases, as compared with 30% in the other diseases (healthy livers excluded) (P < 0.05). Significantly more patients had missed surveillance because of undiagnosed liver disease in those with NAFLD or alcoholic liver disease, as compared with the other diagnoses (P < 0.05). More than 50% of patients with undiagnosed liver disease had an underlying NAFLD or ALD.
The results from the simple and multiple logistic regression analyses are shown in Table 4. In the multiple regression analysis, the diagnosis of NAFLD or alcoholic liver disease was significantly associated with 'No surveillance', having an Odds Ratio of 2.57. Also, the absence of cirrhosis was significantly associated with 'No surveillance'.
Results
Data on the Entire Cohort
In total, 633 patients were registered from January 1st, 2005 to December 31st, 2012. Of these, seven patients were excluded due to incorrect diagnoses and six cases because of treatment received in other hospitals. In four cases exclusion was made due to missing clinical data in the patient files. The remaining 616 patients were included in the study, however, two of which were lost to long-term follow up due to emigration; hence, they were censored at their last clinical contact. Of the 616 patients, 88 patients (14%) had been referred to Karolinska from centres outside Stockholm County ('external referrals'), whereas 528 patients (86%) were residents in Stockholm County. The total number of newly diagnosed HCCs in Stockholm County 2005–2012 was 703 patients; thus, the Karolinska registry covered 75% of all cases diagnosed in the catchment area during this time period.
Patient characteristics are shown in Table 1. Median age at diagnosis was 65.3 years. Of these were 39 (6.3%) <50 years of age, and 52 (8.4%) >80 years of age. One-hundred and forty-seven patients (24%) were female, 468 (76%) were male. Five-hundred and five patients (82%) had cirrhosis, of whom 266 (53%) were Child–Pugh A, 185 (37%) Child–Pugh B, and 44 (8.7%) Child–Pugh C. In 10 cases (2%) Child–Pugh score data were missing. Two-hundred and twenty patients (36%) had diabetes mellitus type 2, and 163 (26%) had the metabolic syndrome. Data for the calculation of BMI was available in 527 (86%) of patients. Median BMI was 26.3. It exceeded 25 in 330 patients (54%) and 30 in 122 patients (20%). Fourteen patients had BMI <19. Median BMI in patients with NAFLD was 30.8, compared to 25.7 in patients with other underlying diagnoses (P < 0.01).
Comparisons Between Periods 1 (2005–2008) and 2 (2009–2012)
As shown in Table 1, the total number of patients increased over time, from 241 in Period 1 to 375 in Period 2, and the number of external referrals increased from 28 to 60 patients. The catchment area coverage was 75% in both Period 1 and 2. The proportion of patients with BCLC-stage B was significantly higher in Period 2, whereas the other BCLC-stages showed no significant change over time. The proportion of Child–Pugh classes was similar between Period 1 and 2. Tumour size and performance status were unchanged, but the number of tumours was significantly higher in Period 2 compared with 1 (3.4 vs. 3.0).
Median BMI increased significantly between time periods, from 25.7 to 26.7 (P < 0.05). The number of patients with NAFLD doubled from 22 in Period 1 to 45 patients in Period 2, increasing the proportion of NAFLD from 9% to 12% of the total cohort. This trend did not reach statistical significance. No significant change was seen among the other diagnoses between Period 1 and 2.
Surveillance Application and Reasons for Missed Surveillance
Surveillance data of the entire cohort and from the two time periods are summarized in Table 1. In short, 134 (22%) of the HCCs were discovered through surveillance. Twenty-eight (4.5%) of these patients were under surveillance according to patient files, but regularity of the investigations could not be controlled as they were referred from other hospitals, and the radiology data were not available. In about half of the surveillance cases, surveillance was incomplete according to our definition.
In 250 patients (41% of the entire cohort), no indication to perform surveillance was found, i.e. there was an absence of either cirrhosis, chronic hepatitis B, or porphyria; or curative therapy for HCC was considered contraindicated, i.e. due to co-existing cardiovascular or pulmonary disease, ongoing abuse or poor general condition. Hence, in these 250 patients the previous decisions not to perform surveillance were considered correct. In 1.1% of cases surveillance was not carried out because of deficient patient adherence. As seen in Table 1, in the remaining 34% of cases, surveillance was indicated but not carried out because of undiagnosed liver disease, undiagnosed cirrhosis, or doctor's failure to order. Of these, 108 patients (18%) had an unknown liver disease, however, in retrospect were considered suitable for surveillance, had their condition been known. Patient records showed that 72 of these 108 patients with unknown liver disease had paid previous visits to a physician under diagnoses such as hypertension, diabetes, alcohol abuse, obesity and elevated liver function tests. Twenty-four patients (3.9%) had a known liver disease (19 of whom had chronic HCV) but undiagnosed cirrhosis. In 79 cases (13%) the cause for not including the patient in a surveillance programme was unclear, and this group was labelled 'Doctor's failure to order'. In 14 of the patients (2.3%) surveillance data was missing. Surveillance patterns were similar in Period 1 and 2.
In three cases the HCCs were missed by ultrasound surveillance, but diagnosed by computerized tomography.
Data on Treatments
The number of resections and local ablative therapies increased in Period 2 compared with Period 1, whereas the number of liver transplantations was constant. The largest increase was seen in palliative treatments in Period 2 compared with Period 1 (P < 0.05). Treatment with TACE increased from 32 (13%) to 90 (24%) and sorafenib increased from 17 (7%) to 52 (14%). There were no significant differences in overall survival, or in survival of the respective treatment groups, between Period 1 and Period 2 (data nor shown) (Table 1).
Treatment and Survival According to Surveillance Data
As seen in Table 2, patients in whom surveillance was indicated but not carried out because of undiagnosed liver disease or cirrhosis, or doctor's failure to order, had significantly larger tumours, a smaller proportion of tumours within UCSF criteria, and they received curative treatment less often as compared with patients who had undergone surveillance (P < 0.05).
There was a trend of larger tumours and smaller proportion within UCSF criteria in patients who underwent incomplete surveillance compared with those receiving complete surveillance; however, the proportion of patients receiving curative treatments was similar (57% vs. 58%) (data not shown).
Survival was significantly better in patients with HCC diagnosed through surveillance compared with no surveillance (Fig. 1A). In the group of patients in whom surveillance was indicated but not carried out because of undiagnosed liver disease or cirrhosis, or doctor's failure to order, survival was worse compared with those who underwent surveillance (Fig. 1B). There was no difference in survival between those who underwent complete vs. incomplete surveillance (data not shown).
(Enlarge Image)
Figure 1.
A. Cumulative proportion surviving (Kaplan–Meier) stratified according to 'Surveillance' (n = 134) or 'No surveillance' (n = 468) (data from the whole patient cohort) (P < 0.05). B. Cumulative proportion surviving (Kaplan–Meier) stratified according to 'Surveillance' (n = 134) or 'No surveillance' only in patients in whom surveillance was indicated (n = 211) (P < 0.05). Here, 'No surveillance' implicates that there was an indication for surveillance but it was not carried out because of undiagnosed liver disease, undiagnosed cirrhosis, or doctor's failure to order.
Clinical Data in Relation to the Underlying Liver Disease
The spectrum of underlying liver disease is demonstrated in Table 3. Forty percent of patients had an underlying hepatitis C, and 17% of these also had a concurrent alcoholic liver disease. Alcoholic liver disease alone was found in 18%, NAFLD in 11% and hepatitis B in 7.5%. Approximately, 10% of patients had a previously healthy liver.
The mean age at diagnosis was highest in patients with NAFLD and in those with previously healthy livers. The metabolic syndrome was significantly more common in patients with NAFLD and alcoholic liver disease as compared with the other underlying etiologies (P < 0.05). The proportion of patients with underlying cirrhosis was lowest in those with NAFLD (healthy livers excluded) (P < 0.05). In patients with NAFLD or alcoholic liver disease, the HCCs were discovered by surveillance in 13% of cases, as compared with 30% in the other diseases (healthy livers excluded) (P < 0.05). Significantly more patients had missed surveillance because of undiagnosed liver disease in those with NAFLD or alcoholic liver disease, as compared with the other diagnoses (P < 0.05). More than 50% of patients with undiagnosed liver disease had an underlying NAFLD or ALD.
The results from the simple and multiple logistic regression analyses are shown in Table 4. In the multiple regression analysis, the diagnosis of NAFLD or alcoholic liver disease was significantly associated with 'No surveillance', having an Odds Ratio of 2.57. Also, the absence of cirrhosis was significantly associated with 'No surveillance'.
Source...