Does the choice of treatment influence survival of patients with small hepatocellular carcinoma in compensated cirrhosis ?
F. Farinati, S. Gianni, G. Marin*, S. Fagiuoli, M. Rinaldi, R. Naccarato.
Cattedra di Gastroenterologia
- Sezione di Gastroenterologia Dipartimento di Scienze Chirurgiche e Gastroenterologiche,
Policlinico Universitario, Padova & *Divisione di Medicina O.C. Dolo, Venezia
patients with small, single hepatocellular carcinoma (HCC) in compensated cirrhosis
are characterized by a relatively good prognosis.
Methods: We report the findings generated in a retrospective study on a cohort of 186 consecutive patients with small (= 5 cm) HCC in Child A or B cirrhosis, who were transplanted (4), underwent surgery (15), or were treated with percutaneous ethanol injection (117), lipiodol chemo-embolization (44) or best supportive care (6), depending on their clinical features.
Results: Overall survival was 26% at 5 years (31% Child A, 20% Child B), with a mean and median survival of 44 and 38 months, respectively. The longest survival was obtained with transplantation and surgery and the worst with best supportive care. When untreated patients were not considered, no significant differences were observed between the different types of treatment, however, even when patients in the Child A group were considered alone. Almost all the patients who underwent surgery relapsed. No significant difference was observed in relation to the stage of the disease, while alpha-fetoprotein levels were singled out as the only relevant prognostic factor in a multivariate Cox's regression model. Costs per year of life saved were extremely high for transplantation and lowest for ethanol injection, with surgery being less expensive than chemoembolization.
Conclusions: This study confirms that patients with single, small HCC nodule in well compensated cirrhosis should be treated and that the choice of type of treatment should be based on the availability of local resources and expertise, on the patients' preference, after they have been properly informed on the survival, morbidity and mortality related to each treatment option, but also on the relative cost of the procedures.
carcinoma, cirrhosis, treatment, survival, cost effectiveness.
Small hepatocellular carcinoma
(HCC) has an unpredictable natural history, with cases showing a constant pattern
of growth, and others either a declining growth rate or a very slow initial
growth followed by a shorter doubling time (1). Several studies have
described the natural course of the tumor in patients with small HCC (less than
5 cm) and, on average, the 3-year survival rate in these patients is in the
13%-30% range, depending on their Child-Pugh status.
Several treatment options are available nowadays for small HCC in cirrhosis, i.e. orthotopic liver transplantation (OLTx) (2), surgical resection (SURG) (3), percutaneous ethanol or acetic acid injection (PEI-PAI) (4,5), radio frequency thermal ablation (RFA) (6) and transcatheter lipiodol-mediated arterial chemo-embolization (TACE) (7). Despite the huge number of studies in the field (8405 papers quoted in Medline), because of the lack of prospective, randomized controlled trials comparing treatment to best supportive care or one specific approach to another (SURG versus OLTx, PEI versus SURG and so on), there is still debate as to whether these patients should undergo surgery or loco-regional treatment. Specifically, the choice between OLTx and SURG may depend on the country where the tumor is diagnosed, the patient's financial resources, the availability of specific surgical expertise, the patient's background and the availability of grafts (8). The comparison between SURG and PEI, on the other hand, has indicated that results are similar, but the available studies compared the two procedures retrospectively and no prospective randomized trials have been published (1). Nonetheless, some authors strongly recommend SURG (9). The situation is further complicated by the recent introduction of RFA (10,11) and microwave or laser techniques (12,13). TACE is used more rarely in patients with solitary, small HCC, but some authors still consider TACE as a very effective treatment in this specific subset of HCC patients (14,15), particularly when the lesion is hypervascular (16).
With this in mind, we attempt to contribute to the debate by describing the results obtained by treating a cohort of 186 consecutive patients with solitary, small HCC in compensated cirrhosis with OLTx, SURG, PEI, TACE or best supportive care (BSC).
From January 1st 1988 to
December 31st 1998, 186 patients affected by solitary, small HCC (< 5 cm)
were consecutively admitted to our departments. Their demographic details are
given in Table
1. These 186 patients make part of a consecutive series of 549 HCC patients
of which they represent 34%; other 186 (34%) had a more advanced disease that
was treated by trans-catheter arterial chemoembolization and 179 32% were not
amenable to any loco-regional treatment and underwent systemic treatment (tamoxifen
or chemotherapy) or only best supportive care.
In compliance with a local policy, the diagnosis was always based on an ultrasound-guided fine needle aspiration biopsy (FNAB) with cytology, micro-histology, or both, obtained using a Toshiba US-SSA ultrasound and a 20-22 gauge Chiba or modified Menghini needle. The cytological material was fixed in air or 80% ethanol and stained with hematoxylin and eosin or Papanicolau. The micro-biopsies were fixed in buffered formaldehyde and stained with hematoxylin and eosin.
No procedure-related complications were observed. The definition of small (< 5 cm) and solitary HCC was always based on the results obtained by at least two imaging techniques (US/CT scan or NMR/spiral CT scan in the last 3 years), but in most patients (all those undergoing OLTx, SURG, TACE and in part of those undergoing PEI) it was also based on the results of a lipiodol angiography, completed by CT scanning after 20-30 days. Most of these patients were recruited through our out-patients clinic, where patients with cirrhosis routinely undergo a 6-monthly follow-up, with alpha-fetoprotein determination and US.
In each patient we recorded the following parameters:
The choice of therapy was made on the basis of the following criteria:
OLTx was preceded in all instances by TACE and followed in 3/4 patients by adjuvant chemotherapy. SURG was based on limited resection (segmentectomy, bisegmentectomy or wedge resection). PEI was carried out in multiple sessions (3 to 12) with 2-8 cc of absolute ethanol injected at each session. In those patients in whom either CT scanning, US Doppler or a control FNAB documented the persistence of viable neoplastic tissue, a second course was performed. TACE was performed with super-selective catheterism of the hepatic artery branch, with 10-15 ml of lipiodol with 20 mg of emulsified epirubicin and final embolization with gelfoam, whenever feasible. A mean 2.8 TACE courses were performed (range 2 to 8). Four TACE courses were performed within the first year of the diagnosis, with subsequent new courses as necessary, i.e. in the event of biochemical or imaging evidence of tumor recurrence/progression after an initial response. TACE treatment was suspended when there were signs of continuous tumor progression despite treatment. Overall, 4 patients underwent OLTx, 15 had SURG, 117 had PEI, 44 had TACE and 6 had BSC. In the 6 cases who received no surgical or loco-regional treatment, the reasons for exclusion were age over 70 with concomitant problems in 2, refusal in 2 and detection of involvement of a portal branch in 2. There was no significant difference in the distribution of patients in the different treatment groups according to the etiology, the only point to underlying being that no patient with alcoholic hepatocellular carcinoma was transplanted.
Following the treatment, the patients were prospectively followed up at our out-patients clinic or in our department, as necessary. The same was true for the small subgroup of patients who received no treatment. Best supportive care was provided to all patients throughout the survival time. In the few cases in which a patient was lost to follow-up, information regarding his status was obtained through the national population registries.
Survival was evaluated by the Kaplan-Meier log rank test (SPSS statistical package), while the relevance of each prognostic factor was evaluated with Cox's forward conditional regression model.
The therapeutic costs were then calculated including the cost of treatment administered to the patients (OLTx, surgery, PEI and TACE), as derived from the Disease-Related Grouping (DRG) system of the Italian Ministry of Health, which scores each therapeutic procedure (OLTx = 28.36 points, surgical resection = 2.9; PEI = 1.2158; TACE = 1.235) and each point is worth 4,850,000 Italian lire, i.e. about US$ 2,420. This approach to the calculation of the therapeutic costs was chosen because of the enormous variability in the actual cost of each procedure, depending on the number and severity of complications and the length of the hospital stay. The calculation specifically includes only the first choice therapeutic procedure and does not take into account adjuvant treatments or any treatment in case of relapse and slightly underestimates the overall costs but this is true for all the treatment modalities. The final figures are $66800 for OLTx, $7020 for surgery, $2930 for PEI and $2983 for TACE. However, TACE was repeated a mean of 2.8 times in our patients, the cost therefore raising to $ 8352. These amounts are what actually the health care system reimburses to any hospital for each treatment-associated admission. All costs were than expressed per year of life saved, by subtracting from the treatment-related the spontaneous survival.
Overall mean and median
survivals were 43.9 and 38 months, respectively, with a 5-year survival of 26%.
No significant difference in survival was found according to gender, age or
etiology of the disease (data not shown). However, when all patients with HCV
or HBV, alone or in association, were compared with those with alcohol abuse
as an etiologic factor, a statistically significant difference emerged: as a
mean, patients with alcohol abuse survived less than those with a viral etiology
or mixed (alcohol plus viruses) etiology (median survival 33.3 months versus
45 and 56 months respectively, and with 5-year survivals of 22, 23 and 46% respectively,
p= 0.012 by log rank test).
Patient subgroups according to Child-Pugh, Okuda, TNM and CLIP stage/score were as follows:
- 116 patients were Child A and 70 were Child B.
- 157 were Okuda I and 29 were Okuda II.
- 47 were T1, 126 were T2, 6 were T3 and 7 were T4.
- 97 were CLIP 0, 74 were CLIP 1 and 15 were CLIP 2.
No significant difference in survival was detected by stratifying the patients according to their Child-Pugh status or Okuda, TNM or CLIP stage. The respective survivals of these staging systems are shown in Table 2 . The survival of patients with HCC in alcoholic cirrhosis was significantly shorter than that of patients with different etiology (Table 1).
The size of the nodule (< or > than 3 cm) had no significant impact on survival, though patients with HCC smaller than 3 cm (83 cases) had a slightly longer mean survival (49 versus 41 months) with a 5-year survival of 29% versus 24%. Nor did the size of the nodule affect the distribution of patients in the different treatment groups (chi squared, p = n.s.).
Segmental portal thrombosis was detected in 11 patients (5.9%) overall, i.e. in 2 untreated and 9 treated patients: in particular in none of the transplanted patients, in 2 SURG , 2 PEI and 5 TACE patients. None of the patients had extra-hepatic secundaries. The presence of segmental thrombosis did not significantly affect patient survival by Kaplan Meier analysis.
The alpha-fetoprotein levels were significantly related to survival (Log Rank 10.3; p=0.01). Patients with normal alpha-fetoprotein levels had a significantly longer survival than patients with even only slightly raised levels (Table 3).
The distribution of patients according to type of treatment and Child-Pugh risk group is shown in Table 4. Patients undergoing PEI were more frequently in the Child-Pugh B group (p=0.009) than in the other subgroups. Mean and median survivals according to type of treatment and 5-year survival (where applicable) are also shown in Table 4. The longest survival was observed after SURG and OLTx, with a 5-year survival of 34% and 50%, respectively.
Overall, 75% of the patients relapsed. In 60% of cases a distant recurrence was observed, in 4% the recurrence was localized at the site of treatment, while in 11% both distant and local relapses were detected. Only 1 out of 4 OLTx patients relapsed, and died of disease 19 months after transplantation, the other death was unrelated with any presence of HCC. Thirteen out of 15 SURG patients relapsed within 5 years of surgery, while one death was observed 15 days after surgery. Only one patient is therefore a non-relapser at 17 months from surgery. All the relapsers underwent TACE. Slightly lower survivals were observed in patients undergoing PEI and TACE as a single treatment, with a 5-year survival of 27% and 15%, respectively. No difference in survival emerged among PEI patients with respect to Child A or B status (median and mean survival 38 and 42.9 for Child A and 33 and 40.4 for Child B, respectively). Since only 4 patients underwent liver transplantation and since SURG and OLTx are both radical procedures - we analyzed survival both considering 5 treatment groups and by subgrouping patients into 4 treatment groups (no treatment, surgical [OLTx+SURG], PEI or TACE). Overall median and mean survivals in the above described group of patients undergoing radical treatment were 58 and 53 months, respectively. The difference in survival was significant both in the first case (p=0.01) and in the latter, with a more strikingly significant difference in this second case (p=0.0023). However, if the patients receiving only BSC are disregarded, no significant difference whatsoever emerged among patients undergoing OLTx, SURG, PEI or TACE with respect to survival (p=0.34).
When all of the above variables were introduced in a Cox's forward conditional regression model, the only variable selected as a prognostic indicator was the alpha-fetoprotein level (p=0.02).
If only the patients with Child A status are considered for the analysis, similar results are obtained with patients undergoing SURG, who survived slightly, but not significantly longer than patients treated with PEI or TACE (median and mean survival 53 and 59.4 for SURG, 38 and 42.7 for PEI, 35 and 39.8 for TACE). Here again, if the untreated patients are disregarded no significant difference whatsoever remains.
Costs per year of life saved were of $21664 for OLTx, 1959 for surgery, 1233 for PEI and 4009 for TACE.
HCC can be considered a lethal disease, but in the last 15 years a dramatic, probably multifactorial improvement in patient survival has been achieved. Several new therapeutic options have been introduced, including liver transplantation, locoregional, US-guided ablative treatments and chemo-embolization. Despite, or probably due to this progress, there is still debate on several issues regarding HCC treatment and one of these certainly concerns the best treatment for "early" HCC in compensated cirrhosis. This is true also in view of the lack of and the difficulty in carrying out prospective randomized trials in these patients, who present peculiar features both from the point of view of their tumor and of the functional status of the underlying cirrhosis. It is important to stress that even our data are not generated in a prospective randomized study, being therefore hardly considerable as sound as those deriving from a prospective randomized trial. However, these data represent the result of the everyday clinical practice in a third level, highly specialized, Gastroenterology Unit and can therefore contribute to the debate. This first of all by showing that, in agreement with previous reports (1, 21 - 23), any treatment is better than best supportive care. Since the study was retrospective, a selection bias with respect to the no-treatment group cannot be ruled out: for instance, 2 of these patients had segmental portal thrombosis, were older, but on the whole they did not differ significantly from the treated patients and survival in this small subgroup was absolutely comparable with cases previously reported in the literature (17% 3-year survival, figure in the range of what reported in the literature).
Apart from the comparison between treatment and BSC, the second fundamental message of this study is that radical treatment, including liver transplantation and surgery, performed only slightly better than PEI or TACE in our series, and the difference was not statistically significant. An apparently improved survival was observed up to month 48, but after 60 months no difference was detectable. Moreover, all the patients who underwent surgery relapsed. The only significant differences in the characteristics of the patients in the different treatment subgroups were a higher percentage of Child-Pugh B patients in the PEI treatment, that was expected, while the distribution according to the etiology was not significantly different, this being an important point since patients with HCC in alcoholic cirrhosis survived less in this experience.
As mentioned previously, the choice among OLTx, SURG and PEI in patients with small, solitary HCC in compensated cirrhosis is still debated. The point has recently been re-addressed by the group in Barcelona, headed by Dr Bruix who suggested that resection is still better than transplantation in carefully-selected cases of early HCC (24). It is interesting to note that, on the basis of the paper by Mazzaferro in the NEJM (2), OLTx is generally considered as the best treatment for patients with early HCC and probably the above-quoted paper will fire a new discussion on the topic. This problem is settled at source in our setting, however, since the chances of a patient with HCC in cirrhosis being transplanted in Italy is somewhat remote due to the shortage of available organs and the long waiting lists involved.
Whether or not PEI should substitute resection, given the higher peri-operative morbidity and mortality related to SURG, the frequent relapses and the similar long-term survival is also debated (25, 26). Again, local policies and the availability of expertise may guide the choice between the two procedures. In our experience, resection performed slightly, but again not significantly, better than PEI, though the number of patients who underwent resection was relatively limited. Again, the comparison may be biased by the fact that the patients in better conditions were selected for surgery. For instance, patients undergoing PEI were more frequently in Child-Pugh B group (though no statistically significant difference was observed in survival according to Child Pugh status among PEI patients) and the nodule's position may have made it easier to treat. In any case, a prospective randomized trial of SURG versus PEI is very difficult to perform, given the huge number of patients required to demonstrate any significant difference and the small share of patients, even among those with early disease, who might really be eligible for randomization.
TACE was apparently the least effective of the invasive procedures in our series. It was selected as definitive treatment when the nodule's position was not amenable to resection or PEI or when the size of the nodule was considered a limiting factor (about 5 cm). None of the above can be considered a priori as the cause of the slightly shorter survival. In fact, the median survival after TACE was initially similar to the situation in PEI-treated patients, but a drop was observed during the last year of follow-up, which led to a 5-year survival rate of 15%. A possible explanation for this finding is that repeated TACE might, in the long term, reduce liver function, causing death due to liver failure and not to tumor progression. In our experience TACE does not impair liver function in the short term (27), but it may become harmful after several repetitions.
This is cohort study and not a prospective randomized trial aimed at comparing the efficacy of different treatments. The patients were allocated to a treatment on the basis of their clinical status or the features of their neoplastic disease but were, thereafter, prospectively followed up. By showing that any significant difference is absent when
the untreated patients are disregarded, this study does however suggest that, whatever the treatment chosen on the basis of a patient's clinical features, the outcome does not differ greatly. Probably, the biological behavior of the tumor is the only relevant variable since the Cox's model only selected the alpha-fetoprotein levels as a prognostic factor. Consequently, the choice of treatment should be based on the morphological and clinical features of the tumor and patient, the local availability of a specific expertise or a favorable setting for liver transplantation and, last but not least, the patients' preference, after they have been exhaustively informed. The information should stress not the prospects of survival, which are similar for all the treatments, but the implications of each type of treatment from the point of view of hospital stay, adverse effects and quality of life, which should be one of the main targets of investigation in such a clinical scenario.
Finally, a relevant issue particularly in countries were resources available for health care are becoming more and more limited is the cost-effectiveness of treatment. There are obvious differences in the costs of providing health care between Italy and other countries, just as there are clear differences in the incidence of HCC and in the therapeutic approach to the disease but this study apparently shows that the less aggressive we are with these patients, the better it is, being the lowest cost per year of life saved associated to PEI.
Faq Links Reviews News E-mail Casi clinici Immagini