Article Text

Surveillance programme of cirrhotic patients for early diagnosis and treatment of hepatocellular carcinoma: a cost effectiveness analysis
1. L Bolondia,
2. S Sofiaa,
3. S Siringoa,
4. S Gaiania,
5. A Casalia,
6. G Zironia,
7. F Piscagliaa,
8. L Gramantieria,
9. M Zanetti,b,
10. M Shermanc
1. aDipartimento di Medicina Interna e Gastroenterologia, Università di Bologna, Italia, bDipartimento di Medicina e Sanità Pubblica, Università di Bologna, Italia, cToronto Hospital (General Division), Canada
1. Dr L Bolondi, Dipartimento di Medicina Interna e Gastroenterologia, Università di Bologna, Policlinico S Orsola-Malpighi, Via Albertoni 15, 40138 Bologna, Italy.bolondi{at}almadns.unibo.it

## Abstract

BACKGROUND Hepatocellular carcinoma (HCC) is a major cause of death in cirrhotic patients. This neoplasm is associated with liver cirrhosis (LC) in more than 90% of cases. Early diagnosis and treatment of HCC are expected to improve survival of patients.

AIMS To assess the cost effectiveness of a surveillance programme of patients with LC for the early diagnosis and treatment of HCC.

PATIENTS A cohort of 313 Italian patients with LC were enrolled in the surveillance programme between March 1989 and November 1991. In the same period, 104 consecutive patients with incidentally detected HCC were referred to our centre and served as a control group.

METHODS Surveillance was based on ultrasonography (US) and α fetoprotein (AFP) determinations repeated at six month intervals. Risk factors for HCC were assessed by multivariate analysis (Cox model). Outcome measures analysed were: (1) number and size of tumours; (2) eligibility for treatment; and (3) survival of patients. Economic issues were: (1) overall cost of surveillance programme; (2) cost per treatable HCC; and (3) cost per year of life saved (if any). Costs were assessed according to charges for procedures at our university hospital.

The cost of treatable HCC was US$17 934 for the surveilled and US$14 555 for the unsurveilled patients (Δ=US$3379) (table 5), resulting in an incremental cost in the surveilled group of US$141 918. Comparison of survival of patients with HCC detected during the surveillance programme (median survival 30 months) with that of patients with HCC detected incidentally in the unsurveilled population (median survival 15 months) showed a cost per year of life saved of US112 996. This estimate, however, does not take into account other incidental costs such as clinic visits, doctors' time, other costs related to hospitalisation, time lost in employment, and travel costs, which further increase the cost of surveillance. Furthermore, it does not take into account the unknown cost in the unsurveilled population (where US and other diagnostic tests, apart from those performed at the time of detection, were randomly performed prior to our observation) whose inclusion in the analysis would improve the cost effectiveness of the programme. ## Discussion Screening and surveillance programmes based on new imaging modalities can produce a cycle of increasing diagnostic and therapeutic intervention without any definite advantage.22 The clinical effectiveness of these programmes relies on the establishment of early diagnosis, provided that effective treatments are available. As a consequence, an increased proportion of successful treatments and a reduction in disease specific mortality of surveilled patients should be the final result. The problem is particularly important in screening for neoplasms arising in diseased organs and in aged patients, as is the case for HCC. This implies that cancer eradication (if possible) would possibly not significantly affect patient survival. Unfortunately, these points have been only partially addressed in previous studies on the problem of screening or surveying cirrhotics for HCC.2-15 23 A randomised study on the potential benefit of surveillance programmes comparing survival in surveilled and unsurveilled patients has never been done and at present it is unrealistic and could even be considered unethical, at least in developed countries where the use of US examinations in patients with liver disease is widespread. In the present study we tried to address this problem, and in addition to the incidence of HCC in a western population we investigated objective measures of effectiveness of a surveillance programme, such as the morphological characteristics of HCC detected during follow up, eligibility for surgical and local treatment, and disease outcome in terms of survival. We then compared these measures with those obtained in HCC incidentally detected outside any specific surveillance programme (non-randomised contemporaneous controls) during the same period of enrollment. Finally, we analysed and reported the costs of the programme in relation to outcome in order to provide data useful for setting priorities in allocating resources.24 The choice of an adequate control group is critical for any study. In our series the volume of unifocal tumours was not significantly different (table 4) between surveilled patients and controls and tumour volume of our controls was smaller than that of other series published recently,25 which included, however, multifocal tumours with clinical symptoms. This fact may support the hypothesis that the same control patients had undergone periodic examinations for their liver disease outside of regular surveillance programmes, a practice that was spreading in Italy at the beginning of the 1990s. On the other hand, the rate of unifocal and diffuse/infiltrating tumours (table 4) between the two groups already favours the efficacy of our surveillance policy, even though the extent of this gain is related to our specific control group. A critical point for the evaluation of cost effectiveness is the proportion of patients with tumours detected during the surveillance programme who are eligible for surgical or local treatment. In our series only 42 of 61 detected cases (68.8%) were treated, a proportion not significantly different from that of the unsurveilled population (58%) who, however, could have been previously selected. Our proportion of treated patients is particularly high if compared with that of Colombo and colleagues9 (only 14% of HCC detected during follow up and 43% of HCC found at entry were operable), Pateron and colleagues12 (28.6%), and Cottone and colleagues13 (33.3%), but is lower than that reported by Oka and colleagues8 (82.5%). This discrepancy is partially apparent if we consider that Colombo et al evaluated only operable cases, while we and Oka and colleagues8 evaluated the various types of local or surgical treatment. Only 6.5% of cases in our study and 17.5% in the study of Oka and colleagues8 underwent hepatic resection. Our cumulative three year survival rate for patients with HCC detected during the surveillance programme was significantly higher (p=2) than that of unsurveilled patients (fig 3) and proved similar (45%) to that reported by Oka and colleagues8 (41%). These values are much higher than those reported in the majority of previous studies on the natural history of unselected series of HCC.25-29 A more recent study30 in patients derived from a prospective randomised trial showed a higher three year survival (50%) but only in patients without adverse factors; this was not the case in our series. Such differences could represent lead time bias or may be due to over diagnosis,22 but this bias seems unavoidable and is further confirmed by the presence of more multifocal disease in unsurveilled patients. The problem of over diagnosis is particularly important from a clinical point of view and surprisingly it has rarely been reported in previous studies on surveillance. If not correctly interpreted, this condition could seriously affect the assessment of outcome measures. In our study, in 13 of 74 detected nodules (17.1%) malignancy was not proved and four of the 61 true HCC were classified as macroregenerative nodules prior to the definite diagnosis or their evolution into HCC. This demonstrates that surveillance discloses quite a large number of nodules of uncertain malignant potential31 causing increasing diagnostic and therapeutic intervention without any demonstration of efficacy, as the natural history of this condition is largely unknown. Another point which has been poorly analysed in previous studies on screening for HCC is cost, and data on this are scarce. Sarasin and colleagues32 recently evaluated the theoretical costs of screening using a decision analysis model but they considered hepatic resection as the only therapeutic option for HCC. Furthermore, even though we believe that theoretical models may have some advantages due to their flexibility, they depend on information derived from different sources and therefore must always be tested and compared with the results obtained in current clinical practice, as in our study. Actual measurement of costs is a major area of methodological inconsistency in cost effectiveness analysis. We adopted the calculation of charges33 because it is the only objective parameter available and comparable in different countries. Therefore, the economic issues reported in this study are based on charges for diagnostic and interventional procedures at our university hospital. As the number of procedures performed is always indicated, readers can easily calculate corresponding charges for their own country. The analysis showed that the cost per year of life saved was US113 534. This does not represent an exact incremental cost effectiveness ratio21 but it offers an idea of the expenditure requested. Cost utility of the programme has not been measured but it is conceivable that screening procedures do not significantly affect quality of life.

Our results demonstrate that a surveillance programme of patients with LC requires a large number of resources and its implementation in clinical practice on a national basis would be questionable in many countries. This conclusion can be understood if we consider that costs for the screening of breast cancer are estimated at US\$13 000–28 000 per year of life saved and still cause debate.34Implementation of any screening programme in the daily practice of each country also requires that such programmes do not consume resources out of proportion of the importance of the disease (prevalence, socioeconomic impact, etc) in the country. Therefore, adoption of screening policies can be different between countries.35However, according to a recent article providing a new perspective in evaluation gains in life expectancy from medical interventions,36 a gain of a year from a preventive intervention targeted at populations with elevated risk similar to what we observed in our study can be considered large.

Finally, it is important to outline that a small proportion of cirrhotic patients seemed to benefit from the surveillance programme (for instance, those transplanted with HCC at a very early stage and not recurring whose life would not have been saved without an early diagnosis). This must stimulate the search for definite individual risk factors or morphological predictors of HCC, such as liver cell dysplasia37 and AgNors quantitation,38allowing for a more targeted surveillance and consequently better cost effectiveness.

## Acknowledgments

This research was supported by grants ex-40% and 60% of MURST (Italian Ministry for Technological and Scientific Research).

View Abstract

## Footnotes

• M Zanetti died on 22 April, 2000.

• Abbreviations used in this paper:
AFP
α fetoprotein
CT
computed tomography
HCC
hepatocellular carcinoma
LC
liver cirrhosis
OLT
orthotopic liver transplantation
PEI
percutaneous ethanol injection
TACE
transarterial chemoembolisation
US
ultrasonography

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.