Correspondence
Correction
Polyprenoic Acid in Hepatocellular Carcinoma
N Engl J Med 1996; 335:1460-1462November 7, 1996
- Article
To the Editor:
We believe that more data are needed before polyprenoic acid can be accepted as an effective drug for the adjuvant treatment of hepatocellular carcinoma, as reported by Muto et al. (June 13 issue).1 We are concerned about several issues that were raised in the report. The authors claimed that tumors recurred in 27 percent of the group given polyprenoic acid (12 of 44) and 49 percent of the placebo group (22 of 45, P = 0.04). However, the Kaplan–Meier analysis in Figure 2 of the article shows a total of 24 patients (16 in the polyprenoic acid group and 8 in the placebo group) who were followed up for less than 24 months. Among the 24 patients, 11 in the polyprenoic acid group and 5 in the placebo group had a follow-up of less than seven months. These 11 patients in the polyprenoic acid group either were not followed long enough or had recurrence of tumor within six months and by definition should not be included in the evaluation of the occurrence of a second primary tumor. In addition, the two-year recurrence rate of 49 percent in the placebo group was not substantiated by the data, since not all patients had reached two years of follow-up.
We are also concerned about the selection of six months as the cutoff point to differentiate a recurrence from a second primary tumor. It is important to know whether this time point was preselected, since a P value below 0.05 can easily be reached if multiple cutoff points are examined.2 Furthermore, sonography was scheduled every three months, but recurrence was noted at seven, eight, and nine months of follow-up. Systemic bias would easily emerge if the investigators were allowed to make a subjective decision about follow-up time. A much more reliable end point is the overall survival rate.
The authors failed to state clearly some statistical points that are pertinent to a phase 3 clinical trial. These include the time at which enrollment was closed, the point at which the results were disclosed after closure of the study, and the schedule for examining results during the trial. If the results were examined at multiple points, the study should be subject to the rule of multiple interim analyses, and a much lower P value would be required to establish statistical significance.3
3 ReferencesChih-Hsin J. Yang, M.D.
Ann-Lii Cheng, M.D., Ph.D.
National Taiwan University Hospital, Taipei, Taiwan 100161
Muto Y ,Moriwaki H ,Ninomiya M , et al. Prevention of second primary tumors by an acyclic retinoid, polyprenoic acid, in patients with hepatocellular carcinoma. N Engl J Med 1996;334:1561-1567
Full Text | Web of Science | Medline2
Altman DG ,Lausen G ,Sauerbrei W ,Schumacher M . Dangers of using “optimal“ cutpoints in the evaluation of prognostic factors. J Natl Cancer Inst 1994;11:829-835
CrossRef | Web of Science3
Pocock SJ ,Hughes MD ,Lee RJ . Statistical problems in the reporting of clinical trials: a survey of three medical journals. N Engl J Med 1987;317:426-432
Full Text | Web of Science | Medline
To the Editor:
In their excellent article, Muto et al. fail to discuss the importance of the profound vitamin A deficiency that affected most patients enrolled in their chemoprevention trial of the synthetic retinoid polyprenoic acid in hepatocellular cancer. The mean (±SD) plasma retinol levels were 8.9±1.1 and 8.7±1.3 μg per deciliter (not per milliliter, as erroneously reported in Table 1 of their article) in the polyprenoic acid and placebo groups, respectively, values that are below the threshold of severe vitamin A deficiency (i.e., 10 μg per deciliter).1 Although these values may be consistent with the expected compromised liver function of this particular cohort, the preventive effect observed in the group receiving the synthetic retinoid may simply have been due to the restoration of normal vitamin A status. It is well known that vitamin A deficiency predisposes patients to alterations in the process of differentiation and to increased cellular proliferation,1 both of which may have enhanced the underlying process of liver carcinogenesis. The observation by Muto et al. seems to confirm the anticarcinogenic effect of retinoid supplementation in subjects with vitamin A deficiency,2 but the broad implication for cancer prevention in subjects with normal vitamin A status remains unclear. Indeed, the recently reported harmful effect of beta carotene supplementation in subjects with normal vitamin A levels would argue against the use of this strategy,3 even though the effects of synthetic retinoids may well be totally different from those of carotenoids.
Another important consequence of vitamin A deficiency is visual problems, particularly night blindness. The authors state that no disorders of visual acuity occurred. However, impaired adaptation to the dark occurs frequently below the threshold of 10 μg of plasma retinol per deciliter, even though it may be unrecognized in some patients.4 Moreover, no mention is made in the article of whether the retinoid modified plasma retinol levels. Other synthetic retinoids significantly decrease plasma retinol levels and induce diminished adaptation to darkness,5 a symptom that may affect the subjects' ability to drive.
5 ReferencesAndrea Decensi, M.D.
Alberto Costa, M.D.
European Institute of Oncology, 20141 Milan, Italy1
Goodman DS . Vitamin A and retinoids in health and disease. N Engl J Med 1984;310:1023-1031
Full Text | Web of Science | Medline2
Li JY ,Taylor PR ,Li B , et al. Nutrition intervention trials in Linxian, China: multiple vitamin/mineral supplementation, cancer incidence, and disease-specific mortality among adults with esophageal dysplasia. J Natl Cancer Inst 1993;85:1492-1498
CrossRef | Web of Science | Medline3
The Alpha-Tocopherol, Beta Carotene Cancer Prevention Study Group
Full Text | Web of Science | Medline4
Russell RM ,Smith VC ,Multack P ,Krill AE ,Rosenberg IH . Dark-adaptation testing for diagnosis of subclinical vitamin-A deficiency and evaluation of therapy. Lancet 1973;2:1161-1164
CrossRef | Web of Science | Medline5
Decensi A ,Torrisi R ,Polizzi A , et al. Effect of the synthetic retinoid fenretinide on dark adaptation and the ocular surface. J Natl Cancer Inst 1994;86:105-110
CrossRef | Web of Science | Medline
Author/Editor Response
The authors reply:
To the Editor: In our study of the prevention of second primary hepatomas by an acyclic retinoid, we included all randomized subjects in the analyses rather than only those who were treated, were eligible, or could be evaluated. This intention-to-treat analysis is the strictest way to evaluate the efficacy of a treatment.1 For example, the first tick mark in the retinoid-treated group in Figure 2 of our article represents a patient who discontinued the drug on the first day because of headache, but who was followed up for another 61 months. The follow-up after randomization ranged from 28 to 64 months for all surviving patients. Withdrawal of patients due to death occurred in the polyprenoic acid group at 12, 13, 21, 25, 30, 32, and 33 months and in the placebo group at 10, 14, 19, 26, 28, 29 (two patients), 31, 32, 37, 39, and 45 months. We are following the remaining patients to determine overall survival rates, since they are the most reliable end points.
The decision to include a cutoff point of six months in the definition of second primary hepatoma was determined before the study. Another definition involving DNA analysis may be possible,2 but we believe our definition is advantageous for a multicenter trial. During the entire period before the study was closed to enrollment in August 1994, no interim analyses were either scheduled or performed. Thanks to the strict protocol and definitions of end points, we did not need to examine the data at multiple points after the study was closed. The data were updated in February 1996 solely to provide the latest follow-up data for a final revision of this article. It is very difficult to avoid a one-month deviation from scheduled examinations in the conduct of a clinical trial like ours.
Ingested retinoid is absorbed in the small intestine, stored in the liver, and delivered to target organs by retinol-binding protein, which is synthesized in the liver.3 The plasma retinol level is determined on the basis of either hepatic retinol stores or the synthesis of retinol-binding protein. In vitamin A deficiency, hepatic retinol stores disappear before the plasma retinol level declines, although retinol-binding protein is available. In contrast, impaired synthesis of retinol-binding protein reduces the plasma retinol level but does not affect hepatic retinol stores in liver cirrhosis. Plasma retinol levels of 8.7 or 8.9 μg per deciliter (not micrograms per milliliter, as was erroneously reported in Table 1 of our article) were consistent with those of patients with hepatoma in our previous study (18.4±14.6 μg per deciliter).4 In such patients, retinoid was not deficient in the noncancerous liver tissue (mean, 318 μg per gram).5 Therefore, the mechanism by which acyclic retinoid inhibited second primary hepatomas in the remnant liver is not related to supplementation of a deficient nutrient. In fact, administration of the retinoid for 12 months did not affect the plasma retinol level in the polyprenoic acid group (8.5±1.2 μg per deciliter). However, in organs other than the liver, symptomatic retinoid deficiency may appear as a result of decreased delivery of retinol by retinol-binding protein in such patients. Sensory disturbances such as impaired adaptation to the dark and diminished taste acuity are typical, as we reported previously.6 However, no sensory disorders were reported by the patients or observed by the investigators in the present study.
6 ReferencesYasutoshi Muto, M.D.
Hisataka Moriwaki, M.D.
Masataka Okuno, M.D.
Gifu University School of Medicine, Gifu 500, Japan1
International Conference on Harmonization of Technical Requirements for Registration of Pharmaceuticals for Human Use. ICH harmonized tripartite guideline: guideline for good clinical performance. Geneva: ICH Secretariat, 1996.
2
Shafritz DA . Synthetic retinoids for the secondary prevention of hepatocellular carcinoma. N Engl J Med 1996;334:1600-1601
Full Text | Web of Science | Medline3
Soprano DR, Blaner WS. Plasma retinol-binding protein. In: Sporn MB, Roberts AB, Goodman DS, eds. The retinoids: biology, chemistry, and medicine. 2nd ed. New York: Raven Press, 1994:257-81.
4
Moriwaki H ,Nomura M ,Okuno M , et al. Clinical assessment of plasma retinol concentration in patients with liver diseases -- with special reference to the appearance of retinyl ester in plasma of patients with severe parenchymal damage. Acta Hepatol Jpn 1985;26:1151-1158
CrossRef5
Muto Y ,Omori M ,Sugawara K . Demonstration of a novel cellular retinol-binding protein, F-type, in hepatocellular carcinoma. Jpn J Cancer Res [Gann] 1979;70:215-222
Medline6
Moriwaki H ,Muto Y ,Ichihara A , et al. Study on gustatory acuity in obstructive jaundice, with special reference to correlation between threshold of taste acuity and plasma retinol-binding protein level after PTC drainage. Acta Hepatol Jpn 1981;22:729-733
CrossRef
