Thymostimulin versus placebo for palliative treatment of locally advanced or metastasised hepatocellular carcinoma: a phase III clinical trial

Patients with histologically proven, locally advanced or metastatic HCC not amenable to or failing established treatment were enrolled; locally advanced tumor was defined as one nodule larger than 5 cm or more than 3 nodules larger than 3 cm in diameter. Pretreatment of the HCC was allowed if tumors had progressed during therapy. However, a treatment-free interval of at least 3 months was required prior to enrollment. Patients between 18 and 80 years of age and a Karnofsky performance status ≥ 60% were eligible. Exclusion criteria were pregnancy/lactation, active second malignancy, severe concomitant disease (e.g. cardiac failure NYHA III-IV, serum creatinine level > 300 μmol/l) or severely decompensated liver function (bilirubin > 5 mg/dl, Child-Pugh > 12). None of the patients received antiviral treatment with interferon during the study period. Ethical approval was obtained from all ethical review boards of the participating centers before study initiation and written informed consent from each patient before entering the study. The study was conducted in accordance with the ethical principles stated in the Declaration of Helsinki and the guidelines on good clinical practice.

The study was designed as a prospective, randomised, controlled, double-blind, multi-centre phase III trial comparing best supportive care plus thymostimulin with best supportive care plus placebo (trial registration: Current Controlled Trials ISRCTN64487365). Patients were recruited at 12 centres in Germany (see Acknowledgements); all biopsies were centrally reviewed by an experienced pathologist (AT). Primary endpoint of the study was 12-month survival, secondary endpoints overall survival, time to progression and tumor response according to standard Response Evaluation Criteria In Solid Tumors (RECIST criteria) [13], toxicity according to ECOG criteria [14] and quality of life assessed by means of the FACT-Hep questionnaire [15] (Figure 1). A sample size of 110 patients was calculated to detect a 20% improvement in one-year-survival in a two-sided log-rank-test with a significance level of 5% and a power of 90%. Allowing for a drop-out rate of at least 20%, 135 patients were planned to be recruited within 24 months. After inclusion of the last patient, a follow-up period of 12 months was scheduled.

Thymostimulin is a licensed immunomodulating drug prepared from an extract of peptides from bovine thymus glands (Thymophysin CytoChemia^® 25/50). Following removal of high-molecular cell components and proteins, the low-molecular active thymus peptides are isolated and standardized to a defined protein fraction. Before central randomization via fax patients were stratified according to liver function (Child classification) and treatment before study entry (received or not received). They were assigned covariate-adapted to each center to either thymostimulin 75 mg or placebo subcutaneously 5 days a week for one year in addition to best supportive care. Study medication (verum and placebo) was coded and labeled to guarantee blinding with the first five injections performed by a medical doctor. Further injections were administrated by the patients themselves or home care workers and registered in a treatment diary. Drug accountability procedures, reporting of (serious) adverse events and source data verification were implemented as part of the trial monitoring according to standard operating procedures conducted by external monitors. Treatment with the study medication was continued for one year or until one of the following criteria was met: disease progression, decompensation of liver function (bilirubin > 5 mg/dl, INR ≥ 2.3) or performance status (Karnofsky score < 50%), unacceptable toxicity, patient refusal or incompliance.

Pre-treatment and follow-up evaluation included a complete medical history, physical examination, blood count and chemistry as well as performance status. Risk factors and severity of liver disease according to Child-Pugh status as well as prior treatment modalities were recorded at baseline. Tumors were assessed by abdominal ultrasound, chest X-ray and either dynamic computerized tomography (CT) or magnetic resonance imaging (MRI). Other imaging techniques such as radionuclide scans were deployed as clinically indicated. Okuda, CLIP and BCLC classifications were used for staging. Follow-up investigations without imaging were conducted every 6 weeks including survival data as well as documentation of concomitant therapies and toxicity of the medication. Tumor response using CT or MRI scanning was evaluated every 3 months according to RECIST criteria [13]: complete response = disappearance of all demonstrable disease; partial response = decrease ≥ 50% of the longest diameter of the target lesion without worsening of all other disease; stable disease = no significant change in disease [decrease < 50% or increase < 25% without new lesions]; progressive disease = increase ≥ 25% of the longest diameter of the target lesion or new lesions). In addition, quality of life assessed by means of the FACT-Hep questionnaire [15] was also measured every 3 months. Follow-up was scheduled for at least one year after enrollment, followed by three-monthly assessments in surviving patients.

All analyses were based on the intention-to-treat (ITT) patient population. Baseline characteristics were expressed as median [range or 95% CI] for continuous variables and percentages (of all patients randomised) for categorical variables. Survival time and time to progression were calculated from the time of randomisation to the date of death or date of progression, respectively. Univariate survival curves for baseline predictors were established with the Kaplan-Meier method and quantitatively expressed as median survival time [95% CI]. As predefined in the study protocol, a stepwise backward Cox's regression analysis of survival was used to adjust the treatment effect according to relative risks imposed by baseline predictors. The following variables were chosen for the first model: age, gender and Karnofsky performance status; the presence of liver cirrhosis and Child classification; Okuda, CLIP and BCLC classification; AFP level, multifocal tumor manifestation, ascites, vascular invasion and extrahepatic metastases; treatment with thymostimulin or placebo and treatment with other therapeutic modalities before study entry. The frequency of adverse events was compared between groups using Chi-square tests. All calculations were done with the SPSS package (version 15.0).

A total of 135 patients were enrolled between October 2002 and March 2005, and randomly assigned to treatment with thymostimulin (n = 67) or placebo (n = 68). Detailed demographic data and tumor-related characteristics are depicted in Table 1. Most patients had liver cirrhosis, and the majority of tumors was staged as advanced HCC (Okuda stage II; CLIP 2-3 points; BCLC stage C) with multinodular tumor growth, vascular invasion and extrahepatic spread. Less than a third of patients had been treated prior to enrollment with surgical resection (R1 or R2 resection), percutaneous ethanol injection (PEI; range 1-5 sessions), transarterial chemoembolisation (TACE; range 1-7 sessions) or systemic hormone- and/or chemotherapy (somatostatin in 7/14 cases; single or combination therapy with tamoxifen, doxorubicin or platin derivates), but suffered from tumor progression. All baseline characteristics were well balanced between the two groups.

Following randomisation, 6 patients in the thymostimulin group and one patient in the placebo group did not receive the allocated treatment (Figure 2). Their data were included in the intention-to-treat analysis; a subgroup analysis excluding these patients did not reveal a different outcome (data not shown). The remaining patients received treatment as scheduled, in most cases until disease progression or deterioration of liver function or performance status. One patient in each group received salvage therapy after tumor progression, in the placebo group TACE, in the thymostimulin group tamoxifen. Secondary tumors requiring alternative therapy were also diagnosed one in each group; in the thymostimulin group, treatment of the patient had to be stopped early due to a hypopharyngeal cancer, in the placebo group the secondary urothelial cancer was detected after 12 months of treatment of the patient. Only 12 patients (thymostimulin, n = 7 [10%]; placebo, n = 5 [7%]) received the study medication for a whole year. At the time of analysis, all patients had stopped their allocated treatment. The median follow-up was 5.2 months [95% CI 4.0-7.3] in the thymostimulin group and 5.6 months [95% CI 3.6-7.3] in the placebo group, the median length of treatment 3.3 months [range 0 to 12 months] for both groups.

For the thymostimulin group, the univariate probability of survival at 6 and 12 months was 42% [95% CI 29-55] and 28% [95% CI 17-41], respectively, with a median survival time of 5.0 months [95% CI 3.7-6.3]. There was no significant difference in univariate survival compared with the placebo group with a 6- and 12-month survival of 40% [95% CI 27-54] and 32% [95% CI 19-44], respectively, and a median survival time of 5.2 months [95% CI 3.5-6.9; p = 0.87; HR = 1.04 [95% CI 0.7-1.6]]. The Kaplan-Meier curve for overall survival is shown in Figure 3A.

The analysis of the objective tumor response showed no complete response in any of the patients. 22 [33%] out of 67 patients in the thymostimulin group achieved stable disease or a partial response (disease-control rate) in the first 6 months followed by progressive disease thereafter. There was, however, a similar tumor response rate in the placebo group in 20 [29%] out of 68 patients, and time to progression did not statistically differ between groups [p = 0.60, HR = 1.13 [95% CI 0.7-1.8]] with a median of 5.3 months [95% CI 2.0-8.6; thymostimulin] and 2.9 months [95% CI 2.6-3.1; placebo], respectively (Figure 4A).

As predefined in the study protocol, an exploratory Cox's regression analysis of survival and time to progression was used to adjust the treatment effect of thymostimulin or placebo according to the relative risks imposed by baseline characteristics. Thus, adjustment for age, gender, Karnofsky performance status, presence of liver cirrhosis and stage of liver disease (Child classification), tumor stage (Okuda, CLIP and BCLC classification), invasive tumor phenotype (vascular invasion or metastases) or treatment prior to enrollment did not reveal a therapeutic effect of thymostimulin in any subgroup of patients. Kaplan-Meier curves for overall survival and time to progression in patients with Child A cirrhosis or tumor stage BCLC B and C are shown in Figure 3B-D and Figure 4B-D, respectively.

Compliance with the allocated treatment was high with an average of 94% and 92% of patients receiving all 5 injections a week in every 6 week-cycle in the thymostimulin and placebo group, respectively (n.s.). While skin reactions following injection are the only known side-effects of thymostimulin [11, 12], it has been shown to reduce chemotherapy-induced toxicity, in particular myelosuppression and infectious episodes [16]. In our study, 219 adverse events were reported - 95 in the placebo and 124 in the thymostimulin group (n.s.). Skin reactions caused the early withdrawal from the trial of two patients on thymostimulin and one patient on placebo (n.s., Table 2). One more patient on placebo withdrew due to adverse events, i.e. diarrhea after injection (Figure 2). 139/219 adverse events were classified as serious - 85 in the placebo and 54 in the thymostimulin group. Using Chi-square-tests, we found a lower incidence of renal failure, ascites and dyspnoea in patients treated with thymostimulin (Table 2). In contrast, hematological or infectious complications were equally common in both groups.

Quality of life was assessed by means of the FACT-Hep questionnaire at baseline and every 6 weeks thereafter. There was no difference between groups in any of the questionnaire's subdomains at any time point.