Interferon alfa therapy is now widely recommended for

Transcription

1 GASTROENTEROLOGY 1999;117: Interferon-Ribavirin for Chronic Hepatitis C With and Without Cirrhosis: Analysis of Individual Patient Data of Six Controlled Trials SOLKO W. SCHALM,* OLA WEILAND, BETTINA E. HANSEN,* MICHELE MILELLA, MING Y. LAI, ANNA HOLLANDER, PETER P. MICHIELSEN, ANTONIO BELLOBUONO, # LILIANA CHEMELLO,** GIUSEPPE PASTORE, DING SHINN CHEN, JOHANNES T. BROUWER,* and the EUROHEP STUDY GROUP FOR VIRAL HEPATITIS *Department of Hepatogastroenterology and Biostatistics, Erasmus University Hospital Rotterdam, Rotterdam, The Netherlands; Department of Infectious Diseases, Karolinska Institutet, Huddinge Hospital, Huddinge, Sweden; Institute of Infectious Diseases, University of Bari, Bari, Italy; Department of Internal Medicine, National Taiwan University Hospital, Taipei, Taiwan; Department of Gastroenterology, University Hospital Antwerpen, Antwerpen, Belgium; # Crespi Division of Medicine and Centre for Liver Diseases, Niguarda Hospital, Milan, Italy; and **Department of Medicine, University of Padova, Padova, Italy Background & Aims: The aim of this study was to compare interferon (IFN)-ribavirin combination therapy with IFN monotherapy in chronic hepatitis C with particular focus on its efficacy in cirrhosis. Methods: A multivariate analysis of individual patient data of all randomized controlled trials using an IFN-ribavirin arm, reported between 1991 and March 1998, was performed. Centers included 1 Asian and 5 European university-based referral centers for liver disease. A total of 197 patients with chronic hepatitis C received IFN- (3 MU three times weekly) and ribavirin (1 1.2 g daily) for 6 months, and 147 patients received IFN- (3 MU three times weekly) for 6 months. Patients were characterized according to previous IFN therapy, presence of cirrhosis, and genotype 1. Efficacy of therapy was evaluated by assessing the sustained response rate by logistic regression analysis. Results: Patients without cirrhosis treated with IFN-ribavirin had a significantly higher sustained response rate than those treated with IFN, approximately 3-fold for previously untreated patients (IFN-ribavirin: genotype 1, 33%; genotype 2/3, 65%; IFN: genotype 1, 8%; genotype 2/3, 24%). In cirrhosis, sustained response rates with IFN-ribavirin (previously untreated: genotype 1, 7%; genotype 2/3, 24%) were also significantly higher than those with IFN (previously untreated: genotype 1, 1%; genotype 2/3, 5%). Clinical relevant superiority of combination therapy over IFN monotherapy was also observed for relapse; the same trend was observed for nonresponders. Tolerance for IFN-ribavirin was similar for patients with or without cirrhosis. Conclusions: Combination with ribavirin significantly enhances the sustained response rate of IFN therapy in major patient types (cirrhosis, genotype 1) with chronic hepatitis C. Thus, IFN-ribavirin combination is likely to become the antiviral therapy of choice for cirrhosis caused by hepatitis C. Interferon alfa therapy is now widely recommended for chronic hepatitis C. 1 However, a sustained response defined by normal serum alanine aminotransferase (ALT) levels and undetectable hepatitis C virus (HCV) RNA 6 months after withdrawal of therapy is observed in only 4% 8% of patients with genotype 1, 2,3 the most prevalent HCV subtype in Europe, North America, and Japan. Similar low response rates have been observed in patients with cirrhosis, 4 who are most in need of effective therapy. Combining ribavirin with interferon alfa enhances the efficacy according to several randomized controlled trials 5 10 and an analysis of 186 individual patient data. 11 The latter report included 78 patients treated with interferon-ribavirin combination. The probability of response was calculated for 6 subgroups based on patient type (previously untreated, relapser, nonresponder) and genotype (1 vs. non-1); the results were only considered to be a crude estimate, and the effect in patients with cirrhosis could not be established. As part of the Concerted Action on Antiviral Therapy of Hepatitis (Eurohep), individual data of an additional 287 patients who were part of 6 randomized controlled trials were assembled. This collection of individual patient data allowed us to analyze the results of interferon-ribavirin combination therapy vs. interferon monotherapy in vari- Abbreviations used in this paper: CRF, case record form; RT-PCR, reverse-transcription polymerase chain reaction by the American Gastroenterological Association /99/$10.00

2 August 1999 INTERFERON RIBAVIRIN IN HEPATITIS C 409 ous patient types of chronic hepatitis C, with particular focus on patients with cirrhosis. Materials and Methods Selection of Participating Centers The principal investigators of randomized controlled studies using an interferon-ribavirin combination arm were identified by screening the abstracts of major liver meetings in Europe till All coordinators invited to participate agreed to cooperate; complete data were usually received when the study was closed by submission or acceptance of a manuscript. Data Collection and Quality Control Approval for the study and the procedures was obtained from the medical ethics committee of the coordination center. Information collected on each patient included pretreatment status (demographics, stage of the liver disease, virus genotype, previous interferon therapy, outcome); treatment data (dose, ALT and HCV RNA, reason for dose reduction or therapy withdrawal); and follow-up data 6 months after therapy (ALT, HCV RNA). In addition, upper and lower limits of normal for the various tests were obtained for each center. Data were submitted by investigators from the participating centers on a 1-page case record form (CRF) per patient. The CRF included definitions of each information item requested to minimize differences in methodology; a separate quality control questionnaire was used for HCV RNA methodology (reverse-transcription polymerase chain reaction [RT-PCR]: sensitivity, quality control; genotype, method used) and for histology scoring (criteria for the diagnosis of cirrhosis). At the coordination center, all data were entered into a temporary database and converted to printouts CRFs. The computer-generated CRFs were checked by the data manager and the study coordinator and returned to the local investigators for correction of missing or questionable data. Transformation of original data to maintain patient privacy (patient s initials were replaced by a code and date of birth by age) and to correct for differences in normal values was done by the biostatistician before the final analysis. Entry and Exclusion Criteria All patients had chronic hepatitis C, defined as anti- HCV positivity for more than 6 months, HCV RNA detectable by RT-PCR, repeatedly abnormal serum ALT levels, and evidence of chronic hepatitis on liver biopsy. The exclusion criteria varied but comprised decompensated cirrhosis and liver disease of other etiology, pregnancy, drug addiction or alcohol abuse, and serious concomitant disease in all studies. Two studies 6,7 excluded patients with previous therapy with interferon and those with cirrhosis; 1 of these studies 6 excluded also females. Patients receiving ribavirin monotherapy or placebo were excluded because they have been described previously. 11 Treatment and Follow-up Combination therapy with interferon and ribavirin comprised 3 MU interferon alfa 3 times per week and mg ribavirin daily in 2 3 doses, both for 6 months. Interferon monotherapy consisted of 3 MU interferon alfa 3 times per week for 6 months. For all patients, follow-up after discontinuation lasted at least 6 months. Outcome Measures The outcome measure for efficacy was the percentage of patients with a sustained response defined by a normal ALT level and undetectable HCV RNA by RT-PCR at end of therapy and at 6 months after therapy, respectively. The outcome measure for tolerability was the percentage of patients with dose reduction or withdrawal of therapy. Statistics Logistic regression analysis 12 was the only method applied for comparison of interferon-ribavirin combination therapy and interferon monotherapy. With this method it is possible to adjust for differences in baseline characteristics between the patient populations in the trials. The trial number is entered into the model as a fixed effect; in this way the effect of treatment or any other variable becomes independent of the trial. A main model containing trial number (1, Milan; 2, Padua; 3, Bari; 4, Rotterdam; 5, Taipei; 6, Huddinge); treatment (interferon-ribavirin, interferon); patient type (previously untreated, relapser, nonresponder); and their interaction was designed (model A). Subsequently, the covariates sex, age, stage of liver disease (cirrhosis, no cirrhosis), and genotype (1, 2/3, other, and unknown) were entered together with potential interaction terms. A backward elimination procedure resulted in model B. Both an intention-to-treat analysis and a per-protocol analysis were performed. Patients with incomplete data at end of therapy or end of follow-up were classified as nonresponders in the intention-to-treat analysis. Sensitivity analyses were performed in three ways: (1) by excluding each study one by one and rerun of models A and B; (2) by a separate analysis of each patient type; and (3) by analyzing each study separately. All results were congruent with models A and B. Hosmer and Lemeshow Goodness-of-Fit test 13 was a standard performed by every analysis to assess the overall model fit. Results are expressed as odds ratios. In addition, an estimate of the probability of a sustained response was calculated for subgroups based on the presence or absence of cirrhosis and

3 410 SCHALM ET AL. GASTROENTEROLOGY Vol. 117, No. 2 genotypes 1 or 2/3 using the principle of weighted means over the different studies. 12 Results Patient Characteristics Of the 344 patients included in the study, 197 underwent interferon-ribavirin combination therapy and 147 interferon monotherapy. The patient characteristics of the 2 groups are shown in Table 1. The groups were comparable for age, sex, ALT level, and genotype; the group receiving interferon-ribavirin combination therapy had a significantly higher percentage of cirrhosis and previous interferon nonresponders. Of the 344 patients in the intention-to-treat population, 338 were evaluable at the end of therapy and 334 at the end of follow-up. The per-protocol population consisted of the intention-to-treat population minus those nonevaluable (n 6), those that had received a wrong treatment schedule (n 1), and those that had stopped treatment (n 20). In total, 27 patients were excluded; the per-protocol population thus consisted of 317 patients. Table 1. Patient Characteristics at Baseline Subgroup All patients (n 344) Interferonribavirin (n 197) Treatment Interferon (n 147) Sex Female 90 (26) 50 (25) 40 (27) Male 254 (74) 147 (75) 107 (73) Age ( yr) Mean (SD) 45 (11) 45 (12) 45 (10) Median a 45 (28 62) 45 (28 62) 45 (28 61) ALT (ULN) Mean (SD) 3.7 (0.1) 3.9 (0.2) 3.6 (0.2) Median a 3.1 ( ) 3.1 ( ) 3.0 ( ) Genotype (58) 114 (58) 87 (59) 2/3 128 (37) 72 (37) 56 (38) Other and unknown 15 (4) 11 (5) 4 (3) Stage of liver disease No cirrhosis 269 (78) 146 (74) 123 (84) Cirrhosis 75 (22) 51 (26) b 24 (16) b Patient type Previously untreated 170 (49) 86 (44) 84 (57) Relapser 60 (17) 30 (15) 30 (20) Nonresponder 114 (33) 81 (41) c 33 (22) c NOTE. Parentheses show percentage of total group falling in specified category. ULN, upper limit of normal. a Fifth and 95th percentiles. b Significant difference, P 0.03 ( 2 test). c Significant difference, P ( 2 test). Response to Therapy Efficacy. The sustained response rate for patients receiving interferon-ribavirin was 56 of 197 (28%) and that for interferon monotherapy was 14 of 147 (9%) patients. When patients with cirrhosis were assessed separately, the sustained response rate after interferonribavirin was 9 of 51 (17%) and that after interferon monotherapy was 0 of 24 (0%) patients. The results of the multivariate analysis for sustained response are shown in Table 2. In the first model (A), no effect of study center was observed, but the effects of treatment and of patient type were highly significant. Patients receiving interferon-ribavirin had a higher chance of responding than those receiving interferon monotherapy. Previous interferon nonresponders had a lower chance of a sustained response than previously untreated patients or relapsers, who did not differ significantly; the effect of interferon-ribavirin in previous interferon nonresponders even failed to remain significant (P 0.16) in a separate analysis of nonresponders only. In the second, advanced model (B), these findings were confirmed and, additionally, genotype was found to be an independent significant factor. Patients with genotype 2/3 responded more frequently than patients with genotype 1. In the advanced model, the presence of cirrhosis tended to be an independent significant factor for previously untreated patients (cirrhotics responding less frequently); the factor cirrhosis was no longer significant for patients with previous interferon therapy. The covariates sex and age had no significant independent predictive value. The result of the per-protocol analysis was similar to the result of intention-to-treat analysis; only the effect of cirrhosis in previously untreated patients was less pronounced. Sensitivity analysis supported the findings of Table 2; the difference in the predicted responses compared with the overall model varied between 3.5% and 7.7% (10th and 90th percentile). The estimated probabilities of a sustained ALT and HCV RNA response derived from the statistical model are shown in Table 3. This statistical model corrects subgroup data for differences in size, baseline characteristics, and trial effect. The percentages derived from the statistical model were within 5% of the actual percentages in all 10 subgroups with more than 10 patients (100%), in 13 of 15 subgroups with 6 or more patients (87%), and in 5 of 9 subgroups with less than 6 patients (55%). Response rates after interferon monotherapy were less than 10% except for patients with genotypes 2 or 3

4 August 1999 INTERFERON RIBAVIRIN IN HEPATITIS C 411 Table 2. Estimated Odds Ratio for Sustained ALT and HCV RNA Response Derived From Multivariate Analysis: Intention-to-treat Analysis (n 344) Covariate Model A a Odds ratio (95% CI) Model B a,b Treatment Interferon-ribavirin vs. interferon c 5.4 ( ) e 5.8 ( ) e Patient type Nonresponder vs. previously untreated 0.07 ( ) e 0.07 ( ) e Relapser vs. previously untreated 0.49 ( ) 0.49 ( ) Genotype 2/3 vs ( ) e Other vs ( ) Stage of liver disease Cirrhosis vs. no cirrhosis d Nonresponder 2.4 ( ) Relapser 1.1 ( ) Previously untreated 0.2 ( ) f a Both models were adjusted for trial effect; P 0.48 and 0.35, respectively (the P values suggest a homogenous treatment effect). b In model B, age (P 0.55) and sex (P 0.34) were dropped by the backward elimination procedure. c The interaction term between treatment and patient type was not significant (P 0.08 and 0.10, respectively). d The interaction term between cirrhosis and patient type was significant: P 0.03 for nonresponders vs. previously untreated. e P 0.001, f P 0.056; P values derived from the log likelihood ratio test. without cirrhosis. For the other 11 subgroups, the sustained rates after interferon-ribavirin were at least 4-fold. In previously untreated patients with cirrhosis, a sustained response rate of 7% (confidence interval [CI], 1% 41%) can be expected for genotype 1, and 24% (CI, 4% 73%) for genotype 2/3. For patients with a relapse after a previous interferon course, the sustained response rate after interferonribavirin appears independent of the presence of cirrhosis: 13% (4% 35%) for genotype 1 and 36% (13% 67%) for genotype 2/3. For nonresponders, the sustained response rate after interferon-ribavirin therapy was also independent of the presence of cirrhosis: 6% (2% 17%) for genotype 1 and 21% (7% 51%) for genotype 2/3 (Table 3). Tolerability. The incidence of adverse events leading to dose reduction or therapy withdrawal is shown in Table 4. Dose reduction and/or therapy withdrawal was significantly more frequent in patients receiving interferon-ribavirin combination (P 0.007, Fisher exact test); the frequency of therapy withdrawal for interferonribavirin combination was 7.6% vs. 3.4% for interferon monotherapy (P 0.11). No significant differences were Table 3. Estimated Percentage of Sustained ALT and HCV RNA Response and 95% Confidence Interval by Patient Type, Genotype, and Stage of Liver Disease Patient type observed in the frequency of dose reduction or therapy withdrawal between patients with or without cirrhosis. The cumulative percentage of patients with hemoglobin levels decreasing to less than 75% the lower limit of normal (e.g., 7.5 g if the lower limit of normal is 10 g), was 5% for patients without or with cirrhosis at 12 weeks; the percentage increased to 7% and 10%, respectively, at 24 weeks. Discussion Interferon-ribavirin Interferon No cirrhosis Cirrhosis No cirrhosis Cirrhosis Previously untreated Genotype 1 33 (18 51) 7 (1 41) 8 (3 17) 1 (0.1 11) 34 (12/35) a 25 (1/4) 7 (3/40) 0 (0/1) Genotype 2/3 65 (47 79) 24 (4 73) 24 (13 41) 5 (0.6 33) 61 (25/41) 0 (0/2) 29 (10/35) 0 (0/4) Relapser Genotype 1 14 (4 35) 10 (2 33) 3 (0.7 10) 2 (0.4 9) 15 (2/13) 0 (0/1) 0 (0/16) 0 (0/1) Genotype 2/3 40 (17 69) 33 (10 66) 9 (2 27) 7 (2 25) 57 (4/7) 33 (3/8) 0 (0/9) 0 (0/4) Nonresponder Genotype 1 5 (1 17) 8 (2 25) 0.5 (0.1 3) 0.7 (0.1 4) 3 (1/35) 8 (2/26) 0 (0/16) 0 (0/13) Genotype 2/3 18 (3 43) 29 (11 58) 1 (0.1 6) 2 (0.3 13) 14 (1/7) 29 (2/7) 0 (0/3) 0 (0/1) NOTE. Estimated percentage derived from multivariate analysis (model B), intention-to-treat analysis. a Data on the second line reflect the percentage of responders per subgroup derived from actual data. The model corrects for differences between trials (center effect) and subgroup size. This study documents a clinically relevant enhanced efficacy of interferon-ribavirin combination therapy over interferon monotherapy in cirrhosis independent of patient type and genotype. Of equal importance are the excellent tolerability of interferon-ribavirin in cirrhosis, Table 4. Tolerance of Treatment in 344 Patients Drug therapy Interferon-ribavirin (n 197) (%) Interferon (n 147) (%) Full therapy course completed Dose reduction 12.1 a 3.5 Stopped 1 or 2 drugs 7.6 b 3.4 a Efficacy seems maintained because the percentage of sustained response was similar to that in the group with full therapy course completed. b Efficacy seems lost because no sustained response was observed in patients who withdrew from therapy.

5 412 SCHALM ET AL. GASTROENTEROLOGY Vol. 117, No. 2 the decrease in hemoglobin level and the withdrawal rate being independent of the presence of cirrhosis. These conclusions are derived from an analysis of individual patient data from 6 randomized controlled trials, all investigator-initiated and performed in different parts of the world. The method of collecting the data was instrumental for obtaining high-quality data. Use of individual patient data made it possible to use one primary outcome measure: ALT level normalization and HCV RNA negativity. No effect of study center on sustained response rate could be detected, whereas marked influences of treatment, patient type, and genotype were readily discernible. These findings provide validity for the observed results despite the variation of methods for HCV RNA analysis, genotyping, diagnosis of cirrhosis, and assessment of side effects in the study centers. The effect of methods used for genotyping (Innolipa [Innogenetics, Ghent, Belgium]; Okamoto, sequencing) was analyzed, but did not change the outcome (data not shown). Responses to interferon-ribavirin combination therapy in patients with cirrhosis now appear to be similar to those obtained with interferon monotherapy during 12 months in noncirrhotic patients with chronic hepatitis C. 1 In the 1997 consensus report on therapy of chronic hepatitis C, 1 interferon monotherapy for cirrhosis caused by hepatitis C was not recommended outside controlled studies. The outcome of this study, however, suggests that clinically relevant response rates are obtainable with combination therapy in cirrhotics, in particular for patients with genotypes 2 and 3. The outcome of this study should be viewed against those from recently published large randomized controlled trials The large randomized controlled trial is generally viewed as the gold standard. However, it is very costly and the study population is usually only part of the spectrum of the disease that is in need of treatment. Because a large trial is often designed on the basis of small promising randomized controlled trials, it is important for medical practice to explore whether the combination of small trials can yield results of similar validity as those from large trials. Classical meta-analysis has been proposed but also criticized recently. 16 We believe that meta-analysis of individual patient data if performed properly can yield results of similar validity to large trials; moreover, as in this study, it can provide evidence that the observed effects are applicable for a spectrum of disease that is wider than that of the study population of the large trials. In this study we provide evidence that patients with cirrhosis also have a clinical relevant benefit from interferon-ribavirin combination therapy. References 1. Anonymous. National Institutes of Health Consensus Development Conference Panel statement: management of hepatitis C. Hepatology 1997;26(suppl 1):2S 10S. 2. Brouwer JT, Nevens F, Kleter B, Elewaut A, Adler M, Brenard R, Chamuleau RAFM, Michielsen PP, Pirotte J, Hautekeete ML, Weber J, Bourgeois N, Hansen BE, Bronkhorst CM, Heijtink RA, Fevery J, Schalm SW. Efficacy of interferon dose and prediction of response in chronic hepatitis C: Benelux study in 336 patients. J Hepatol 1998;28: Keefe EB, Hollinger FB, the Consensus Interferon Study Group. Therapy of hepatitis C: consensus interferon trials. Hepatology 1997;26(suppl 1):101S 107S. 4. Schalm SW, Fattovich G, Brouwer JT. Therapy of hepatitis C: patients with cirrhosis. Hepatology 1997;26(suppl 1):128S 132S. 5. Brouwer JT, Nevens F, Michielsen P, Hautekeete ML, Chamuleau RAF, Adler M, Delwaide J, Heijtink RA, Schalm SW. What options are left when hepatitis C does not respond to interferon? Placebo-controlled Benelux multicentre trial on ribavirin monotherapy versus prolonged combination with Interferon. J Hepatol 1994;21:S Chemello L, Cavaletto L, Bernardinello E, Guido M, Pontisso P, Alberti A. The effect of interferon alfa and ribavirin combination therapy in naive patients with chronic hepatitis C. J Hepatol 1995;23(suppl 2): Lai MY, Kao JH, Yang PM, Wang JT, Chen PJ, Chan KW, Chu JS, Chen DS. Long-term efficacy of ribavirin plus interferon alfa in the treatment of chronic hepatitis C. Gastroenterology 1996;111: Bellobuono A, Mondazzi L, Tempini S, Silini E, Vicari F, Ideo G. Ribavirin and lymphblastoid -interferon vs -interferon alone in the retreatment of chronic hepatitis C: a randomized clinical trial. J Viral Hepat 1997;4: Reichard O, Norkrans G, Fryden A, Braconier J-H, Sönnerborg A, Weiland O. Randomised, double-blind, placebo-controlled trial of interferon -2b with and without ribavirin for chronic hepatitis C. Lancet 1998;351: Milella M, Santantonio T, Pietromatera G, Maselli R, Casalino C, Appice A, Mariano N, Genchi C, Pastore G. Ribavirin plus IFN vs IFN alone in the treatment of either non-responder or relapser patients with chronic hepatitis C. International Viral Hepatitis and Liver Disease Symposium, Rome, Schalm SW, Hansen BE, Chemello L, Bellobuono A, Brouwer JT, Weiland O, Cavalletto L, Schvarcz R, Ideo G, Alberti A. Ribavirin enhances the efficacy but not the adverse effects of interferon in chronic hepatitis C. Meta-analysis of individual patient data from European centres. J Hepatol 1997;26: Hosmer DW, Lemeshow S. Applied logistic regression. 1st ed. New York: Wiley, Poynard T, Marcellin P, Lee SS, Niederau C, Minuk GS, Ideo G, Bain V, Heathcote J, Zeuzem S, Trepo C, Albrecht J. Randomised trial of interferon 2b plus ribavirin for 48 weeks or for 24 weeks versus interferon 2b plus placebo for 48 weeks for treatment of chronic infection with hepatitis C virus. Lancet 1998;352: McHutchison JG, Gordon SC, Schiff ER, Shiffman ML, Lee WM, Rustgi VK, Goodman ZD, Ling M-H, Cort S, Albrecht JK. Interferon alfa-2b alone or in combination with ribavirin as initial treatment for chronic hepatitis C. N Engl J Med 1998;339: Davis GL, Esteban-Mur R, Rustgi VK, Hoefs J, Gordon SC, Trepo C,

6 August 1999 INTERFERON RIBAVIRIN IN HEPATITIS C 413 Shiffman ML, Zeuzem S, Craxi A, Ling M-H, Albrecht JK. Interferon alfa-2b alone or in combination with ribavirin for the treatment of relapse of chronic hepatitis C. N Engl J Med 1998;339: Lau J, Ioannidis JPA, Schmid CH. Meta-analysis duet. Summing up evidence: one answer is not always enough. Lancet 1998;351: Received July 21, Accepted April 27, Address requests for reprints to: Solko W. Schalm, M.D., Department of Hepatogastroenterology, University Hospital Rotterdam Dijkzigt, P.O. Box 2040, 3000 CA Rotterdam, The Netherlands. devlaming@inw2.azr.nl; fax: (31) The authors thank many individuals for constructive help in the Eurohep Group studies. Clinicians who allowed the use of individual patient data are as follows. Sweden: J. H. Braconnier, A. Fryden, G. Norkrans, O. Reichard. Benelux: M. Adler, R. A. F. M. Chamuleau, J. Delwaide, A. Elewaut, J. Fevery, M. L. Hautekeete, F. Nevens, J. W. den Ouden-Muller, P. J. Wismans. Bari: C. Casalino, R. Maselli, G. Pietromatera. Taipei: K. W. Chan, J. H. Kao. Milan: G. Ideo. Padua: A. Alberti, L. Cavalletto. Data collection and checking was done by J. Boot and C. van Vliet (Rotterdam).