Hemochromatosis genes and other factors contributing to the pathogenesis of porphyria cutanea tarda

All patients with PCT, controls with hemochromatosis, and normal controls were evaluated at the General Clinical Research Center under a protocol approved by the University of Utah Institutional Review Board. Both groups of controls were members of hemochromatosis pedigrees previously reported.19 20 All study participants gave written informed consent. Subjects who agreed to undergo a liver biopsy received a careful explanation of the risks associated with this procedure and provided specific written informed consent.

The diagnosis of PCT was based on the presence of a characteristic bullous dermatosis associated with skin fragility and an increase in the 24-hour urinary excretion of uroporphyrin and 7-carboxyl porphyrin. The diagnosis of F-PCT was assigned to patients with approximately half-normal activity of URO-D measured in erythrocyte lysates by a modification of the method of Straka et al.21 22 Control samples from normal subjects were included in each assay. The diagnosis of S-PCT was assigned when values did not differ from controls.

Urine porphyrins were quantified spectrophotometrically3and by high-performance liquid chromatography (HPLC).21 23 

The serum iron concentration and transferrin saturation were measured as previously described.24 When possible, samples were obtained after subjects had fasted overnight. Serum ferritin concentrations were measured with radioimmunoassay kits (Ramco Laboratories, Houston, TX).

Hepatic iron stores were assessed by light microscopy (graded on a scale of 0 to 4) according to the method of Scheuer et al.25 and by atomic absorption spectrophotometry.24 The normal value for hepatocellular stainable iron is grade 0 to 1. Normal values for hepatic iron concentration are less than 25 μmol/g dry weight.19Histologic grading and staging of HCV was performed by the method of Batts and Ludwig.26 Grading is a measure of the severity of the necroinflammatory process, and staging refers to the degree of fibrosis present.

Anti-HVC IgG (immunoglobulin G) was detected serologically using a commercially available enzyme-linked immunoabsorbent assay (ELISA 2.0, Chiron, Emeryville, CA). A positive result is designated HCV+ and a negative result HCV−.

DNA was available from 87 patients with PCT (64 with S-PCT and 23 with F-PCT) and from 56 clinically normal subjects married to members of well-characterized hemochromatosis pedigrees.19,20 The ancestral cysteine 282 tyrosine (Cys282Tyr) HFEmutation and the histidine 63 asparagine (His63Asp) mutation were detected using polymerase chain reaction (PCR) amplification as described by Feder et al.27 

Alcohol consumption was estimated with a questionnaire and graded as follows: grade 0: nondrinkers; grade 1: consumption of less than 20 g of alcohol per day; grade 2: consumption of between 20 g and 70 g of alcohol per day for a minimum of 3 consecutive years; grade 3: consumption of more than 70 g of alcohol per day for a minimum of 3 consecutive years.

Statistical analyses were performed using the Prophet computer software package.28 Serum iron, transferrin saturation, and serum ferritin concentration values in PCT patients were compared to age- and sex-matched controls with the Wilcoxon signed rank test. Frequencies of the HFE genotypes were compared with the Fisher exact test. To calculate an odds ratio for the genotype Cys282Tyr/Cys282Tyr as a risk factor for PCT, we assumed the frequency of the homozygous mutant genotype among controls to be 0.005.27,29 The Mantel-Haenszel test was used to calculate an odds ratio for alcohol, HCV, and estrogens in PCT patients with the Cys282Tyr/Cys282Tyrgenotype. We employed age- and sex-matched controls with hemochromatosis but without PCT who were members of well-characterized hemochromatosis pedigrees.19 We used 3 controls matched by age (plus or minus 5 years) and sex for each patient with PCT.

We evaluated 108 patients with PCT from 1977 to 1998. Of these patients, 31 (29%) had F-PCT and 77 (71%) had S-PCT. The F-PCT patients were members of 30 pedigrees. There were 18 males (mean age, 35 years) and 13 females (mean age, 45 years); 3 F-PCT patients were children: 2 brothers aged 6 and 10 years and an unrelated 12-year-old girl. The 2 clinically affected young brothers also had a clinically affected paternal uncle (not studied by us). This was the only F-PCT pedigree we evaluated with more than 1 clinically affected member. Of the 77 patients with S-PCT, 50 were males (mean age, 43 years), and 27 were females (mean age, 43 years). We found 2 siblings with S-PCT, a 53-year-old woman and her 49-year-old brother. Both also had hemochromatosis. Both alleles of URO-D were sequenced in these siblings, and no mutations were identified (data not shown). One female S-PCT patient was 5 years old.

The mean 24-hour urinary secretion of porphyrins in the 29 patients with F-PCT was 4061 μg (range, 749-9376 μg/24 h). The mean value for 24-hour urine porphyrin secretion in the 77 patients with S-PCT was 4932 μg (range, 730-26 921 μg/24 h). The highest 24-hour urine porphyrin excretion, 26 921 μg, was found in a 65-year-old woman who was receiving chemotherapy for metastatic lung cancer. The urine porphyrin profile in both F-PCT and S-PCT was similar, with uroporphyrin and 7-carboxyl porphyrin predominating. Normal subjects studied in our laboratory secrete less than 200 μg/24 h of urinary porphyrins, and the predominant urinary porphyrin is coproporphyrin.

Of the 108 patients with PCT, 9 had initiated phlebotomy therapy prior to our evaluation. The iron phenotype of the remaining 99 patients and 99 controls matched by age and sex is shown in Table1. No significant differences in the iron phenotype were detected when patients with S-PCT were compared to patients with F-PCT. Values for the serum iron, transferrin saturation, and serum ferritin concentration in patients with PCT were significantly higher than in controls (Table 1). Hepatic parenchymal cell stainable iron was graded on 89 liver biopsies (60 obtained from males and 29 from females). Only 1 patient had no histologic evidence of hepatocellular iron stores (grade 0), yet her porphyria improved with phlebotomy therapy. This patient, a 27-year-old nulliparous woman with active systemic lupus erythematosus, was ingesting birth control pills at the time PCT developed. She did not drink alcohol and had no evidence of HCV. Liver biopsies from the remaining patients were graded for iron as follows: grade 1: 30 patients; grade 2: 37 patients; grade 3: 14 patients; and grade 4: 7 patients. The hepatic iron concentration measured by atomic absorption spectrophotometry was normal in the single subject with grade 0 liver iron and in 10 of the 30 patients with grade 1 liver iron.

HFE genotypes for 87 patients with PCT are shown in Table2; 19 patients were Cys282Tyrhomozygotes. Only 17 of these patients are shown in the table, as 2 pairs of Cys282Tyr homozygous siblings were evaluated; 1 sibling pair had F-PCT, and 1 pair had S-PCT. Homozygosity for theCys282Tyr mutation was clearly overrepresented. The controls had the same frequency of heterozygosity for the Cys282Tyrmutation. Compound heterozygotes for the Cys282Tyr/His63Aspmutations also were overrepresented, but the number of subjects with this genotype was small. Of the 87 PCT patients included in Table 2, 174 chromosomes are represented. The Cys282Tyr mutation was carried in 53 (30%) of these 174 chromosomes compared with 7 (6%) out of 112 chromosomes (6%) in 56 nonporphyric Utah controls (P < .0001).

We also compared the frequency of the His63Asp mutation in these groups. For this comparison all chromosomes bearing theCys282Tyr mutation were removed from the calculation becauseHis63Asp mutations have not been detected on chromosomes with the Cys282Tyr mutation.30 Of the remaining 121 chromosomes from PCT patients, 31 (26%) carried the His63Aspmutation compared with 12 (11%) out of 105 chromosomes in the controls (P = .007). The frequency of the Cys282Tyr mutation on chromosomes from patients with S-PCT (45/132 [34%]) did not differ significantly from the frequency on F-PCT chromosomes (12/46 [26%]), but the frequency of His63Asp was significantly higher on chromosomes from F-PCT (13/46 [28%]) than on chromosomes from S-PCT (18/132 [14%]) (P = .02).

The hepatic iron concentration, transferrin saturation, and the serum ferritin concentration were highest in PCT patients who were homozygous for the Cys282Tyr mutation (Table3). Regardless of the HFE genotype, patients with PCT had a higher transferrin saturation and serum ferritin concentration than the controls. The mean liver iron grade and transferrin saturation associated with compound heterozygosity at theHFE locus (Cys282Tyr/His63Asp) were higher than those in simple heterozygotes (Cys282Tyr/WT), but statistical significance could not be established because of the small number of compound heterozygotes.

Serologic testing for anti-HCV IgG was performed on 71 patients with PCT. There were 42 (59%) seropositive (HCV+) patients. Males were more likely to be HCV+ (36/45 [80%]) than females (6/26 [23%]). There was no significant difference in the frequency of HCV+ between patients with S-PCT (35/56 [63%]) and those with F-PCT (7/15 [47%]) (P = .3). Of the 42 HCV+ patients, 27 (64%) had used intravenous drugs in the past, and 1 patient had received blood transfusions following an automobile accident 17 years before PCT developed. There was no significant difference in the frequency of HCV+ in PCT patients with any HFE genotype, nor was the iron phenotype influenced by HCV status.

Approximately 1.8% of the population of the United States has antibodies against HCV.31 We tested 64 PCT patients for both the anti-HCV antibody and the Cys282Tyr mutation. We screened 14 PCT patients who were homozygous for the Cys282Tyrmutation for anti-HCV IgG, and 6 (43%) were HCV+. We also tested 60 nonporphyric hemochromatosis homozygotes, and only 3 (5%) were HCV+. These data clearly indicate that the prevalence of HCV+ in patients with PCT greatly exceeds the prevalence in the general population. The odds ratio of developing PCT in hemochromatosis homozygotes is 14 when HCV antibody is present versus when it is not.

We were able to grade 84 of the 89 liver biopsies for active inflammation and staging of fibrosis. Biopsy samples were taken from 39 HCV+ patients, 20 HCV− patients, and 25 patients not tested for HCV. Results are shown in Table 4. Bridging fibrosis (stage 3) and cirrhosis (stage 4) were most prevalent in HCV+ patients, particularly in heavily iron-loaded patients with PCT who consumed over 70 g of alcohol daily. The effect of theCys282Tyr HFE mutation on the degree of fibrosis in HCV+ patients was also analyzed. We genotyped 58 chromosomes in patients with stage 0 to 2 fibrosis, and 26% carried theCys282Tyr mutation. A higher proportion ofCys282Tyr–bearing chromosomes was detected in patients with grade 3-4 fibrosis (6 of 12 genotyped chromosomes [50%]), but this difference did not reach statistical significance (P = .09).

Average daily alcohol consumption was estimated in all 108 patients as follows: grade 0 (consumed no alcohol): 11 (10%) patients, including 3 children; grade 1 (drank only occasionally): 24 (22%) patients (5 men and 19 women); grade 2 (consumed between 20 and 70 g of alcohol daily): 23 (21%) patients (17 men and 6 women); and grade 3 (consumed more than 70 g of alcohol daily): 40 (46%) patients (42 men and 8 women). We found 46 patients with PCT who did not have theCys282Tyr mutation at either HFE allele (Table 3); 27 of these patients consumed more than 70 g of alcohol daily (grade 3). Neither the ferritin concentration nor the liver iron content in this group of 27 patients differed significantly when compared to patients who were heterozygous for the Cys282Tyr mutation and who were also heavy drinkers (grade 3).

Of the 38 adult women with PCT, 24 (63%) were using oral estrogen preparations when PCT was first detected. None were using transdermal estrogen delivery systems. Of the 13 premenopausal women, 9 (60%; mean age, 33 years plus or minus 9 years) had been taking oral contraceptives for more than 6 months (range, 0.5-15 years). Of the 25 postmenopausal women, 15 (60%) were taking oral estrogen replacement therapy for more than 1 year (range, 1-20 years). In the premenopausal group, there were 7 women with S-PCT and 2 with F-PCT. In the postmenopausal group, there were 10 women with S-PCT and 5 with F-PCT. Estrogen ingestion was the only identified factor known to be associated with PCT in 9 women (7 with S-PCT and 2 with F-PCT). Five of these women had prominent hepatic steatosis; 1 had estrogen-induced hepatic adenomas, but the porphyrin content of a biopsy specimen taken from the adenoma did not differ from a sample from a normal portion of the liver (data not shown). We compared serum iron concentration, transferrin saturation, serum ferritin concentration, liver iron grade, and HCV status between women with PCT who were taking estrogens and those who were not. No significant differences were noted (data not shown).

The Cys282Tyr/Cys282Tyr genotype was clearly a risk factor for the development of PCT (odds ratio of 60; 95% confidence limits (CLs), 18.5 and 195.7). To estimate the role of risk factors, such as HCV, alcohol, and estrogens, in Cys282Tyr homozygotes, we compared Cys282Tyr homozygotes with PCT to age- and sex-matchedCys282Tyr homozygotes without PCT (Table5). Homozygotes for the Cys282Tyrmutation with grade 3 alcohol consumption had higher ferritin values and liver iron content than alcohol abusers who were not homozygotes.

We attempted to evaluate HCV and alcohol as independent risk factors inCys282Tyr homozygotes, but these 2 factors proved to be highly associated. All HCV+ homozygotes had grade 2-3 alcohol consumption, and only 2 alcohol users were HCV−. Estrogen use, however, did prove to be a risk factor independent of HCV status. Estrogen use inCys282Tyr homozygous women without HCV was compared to matched controls without HCV, and the odds ratio for estrogen as a risk factor remained high at 9.0 (95% CLs, 1.1-71).

The analysis was expanded to PCT patients of all HFE genotypes, and again HCV and alcohol were highly associated. Of the PCT patients with HCV, 34/36 men and 6/6 women reported grade 2-3 alcohol consumption. Only 9 PCT patients with grade 2-3 alcohol consumption did not have HCV; 2 were Cys282Tyr homozygotes, 2 had F-PCT, and 1 had hepatitis B virus. In this study, 8 women without HCV and with only grade 0-1 alcohol consumption were taking estrogens, thereby confirming estrogen use as an independent risk factor. Only 1/108 PCT patients we evaluated, a 5-year-old girl with S-PCT, had no risk factors and was genetically normal at the URO-D locus.

Most patients with PCT had multiple risk factors; 80% of the men had more than 1 risk factor, and the most common combination was alcohol and HCV, which was found in 71% of these patients. Single risk factors were more common (40%) in women because of the effect of estrogen use. In 28% of women with PCT, estrogen was the only risk factor identified, and in 12% of the women, either alcohol or theCys282Tyr/Cys282Tyr genotype was the only risk factor identified. More than 1 risk factor was found in the remaining 60% of women with PCT, and the most frequent combination was alcohol and HCV. This combination was found in 28% of the PCT women we studied.

Mutations at the HFE locus, HCV infection, excess alcohol intake, and exposure to estrogens all proved to be risk factors for the development of PCT. HFE mutations and HCV infection imparted the greatest relative risk, and the presence of multiple risk factors was more frequent than the presence of single risk factors.

Hepatic iron overload is a nearly constant finding in patients with PCT, and iron depletion is the mechanism underlying the beneficial effect of phlebotomy therapy.14 Most patients with clinically evident hemochromatosis are homozygous for the ancestralCys282Tyr mutation at the HFE locus,27 and it has been estimated that approximately half of the Cys282Tyrhomozygotes develop clinically significant iron overload.32Heterozygotes develop laboratory evidence of iron overload in about 20% of cases.19 A second mutation, designatedHis63Asp, is found in the heterozygous state in approximately 20% of the Caucasian population.27 The clinical consequences of the His63Asp mutation are probably minor, although compound heterozygosity for the 2 mutations (genotypeCys282Tyr/His63Asp) has a greater effect on body iron burden than simple heterozygosity for the Cys282Tyr mutation (genotypeCys282Tyr/WT).30 

Mutant HFE alleles were detected in 63% of the PCT patients we genotyped (Table 2). Homozygosity for the Cys282Tyr mutation was found in 19%, and an additional 15% were simple heterozygotes (genotype Cys282Tyr/WT). These results confirm the report of Roberts et al.,33 who found that 17% of 41 British patients were Cys282Tyr homozygotes and 20% were heterozygotes. Similar findings were also reported by Bonkovsky et al18 in 26 American PCT patients. Collectively, these data suggest that the Cys282Tyr/WT genotype is not overrepresented in patients with PCT. All of our patients with PCT had an iron phenotype that differed significantly from nonporphyric controls (Table3), but only Cys282Tyr homozygotes had an iron phenotype different from PCT patients with any other HFE genotype. These data indicate that the most dramatic effect of HFE mutations on the hepatic iron overload seen in PCT occurs when theCys282Tyr/Cys282Tyr genotype is present.

We found 6 patients with the Cys282Tyr/His63Asp genotype, but we could not confirm an effect on the iron phenotype (Table 3). In a study of Italian men with PCT, Sampietro et al34 found heterozygosity for the His63Asp mutation (genotypeHis63Asp/WT) more often than in controls, but the mutation was not related to iron status. The His63Asp/WT genotype was not over represented in our PCT patients (Table 2), and we also found no impact of this genotype on the iron phenotype (Table 3).

The mechanism responsible for liver iron loading in PCT patients without a Cys282Tyr or His63Asp mutation has not been defined. Other HFE mutations associated with iron overload have recently been described, but these occur infrequently.35,36The HFE gene product regulates cellular iron metabolism through an interaction with the membrane-bound transferrin receptor,37 but whether this interaction enhances or diminishes delivery of iron to cells remains unresolved.38 39 

Both HCV infection and alcohol excess have been associated with hepatic siderosis. We found no significant difference between the frequency of HCV infections in PCT patients with mutant HFE alleles and those with WT alleles, which suggests that HFEmutations do not predispose to HCV infections. A similar conclusion was reached by Smith et al40 in a study of 137 British patients with chronic HCV infections. Our data, however, clearly indicate that HCV infection predisposes to the development of PCT, confirming earlier reports. Approximately 75% of patients with PCT in the United States, Italy, Spain, and France have laboratory evidence of HCV exposure,18,41,42 but the HCV association is less common in patients with PCT from Northern Europe, Australia, and New Zealand.43 44 

An increase in the serum iron concentration, the percent of saturation of transferrin, and the serum ferritin concentration occurs commonly in patients with HCV,45,46 although hepatic siderosis is a less common occurrence. When the serum ferritin and hepatic iron concentrations are elevated, HCV patients are less likely to respond to therapy with α interferon (IFN-α).47 It remains unknown, however, if infection with HCV leads to hepatic iron excess or if preexisting siderosis enhances the damaging effects of the virus. All of our patients with PCT and HCV responded to phlebotomy therapy, but IFN-α therapy has proven useful in the rare instance when phlebotomy therapy has failed.48 

We found the most severe changes in liver biopsy samples from PCT patients with HCV who were iron loaded and consumed more than 70 g of alcohol daily. Many reports have indicated that alcohol accelerates the development of hepatic fibrosis in patients with HCV.49,50The association between excess alcohol consumption and PCT is well recognized,11 but we found it difficult to evaluate alcohol as an independent risk factor because both alcohol abuse and HCV were found in 71% of our male patients and 28% of our female patients. Animal studies have demonstrated that alcohol intake causes hepatic iron deposition,51 and it has been estimated that up to one-third of alcoholics develop hepatic iron overload.52LeSage et al53 concluded that alcohol-associated iron overload was likely due to the effects of human leukocyte antigen–associated (HLA-associated) hemochromatosis, and Simon et al54 confirmed this conclusion. Our data suggest that alcohol is associated with hepatic iron loading in patients with PCT, but this effect is magnified by homozygosity for the Cys282Tyr HFE mutation. Phlebotomy therapy is effective in treating PCT in alcoholics, even if excessive alcohol intake continues throughout therapy.14 This strongly suggests that iron is the key etiologic factor.

Approximately 18% of American women between the ages of 18 and 44 use oral contraceptives,55 and 15% of postmenopausal American women use medicinal estrogens as hormone replacement therapy.56 Of the women with PCT in our study, 63% were ingesting estrogen preparations at the time that PCT became clinically apparent, and in 28% of these women, estrogen was the only precipitating factor identified. Estrogen proved to be an independent risk factor for the development of PCT, but the mechanism responsible for the estrogen effect is unknown. Estrogens have been reported to cause hepatic steatosis, intrahepatic cholestasis, hepatic adenomas, hepatocellular carcinomas, and peliosis hepatitis,57-60 but hepatic siderosis has not been recognized. Clinical and biochemical remission of PCT without phlebotomy therapy has been reported in women who terminate the use of estrogens, but this occurs only when exposure to estrogens has not been prolonged.61 

The activity of hepatic URO-D is reduced in both F-PCT and S-PCT. Half-normal activity of hepatic URO-D in heterozygotes forURO-D mutations is not rate limiting, as most carriers of mutant alleles of URO-D do not express a clinical phenotype.62 In S-PCT hepatic URO-D, protein concentration is normal even though enzyme activity is reduced. Our data indicate that the same risk factors (iron excess, alcohol, HCV, and estrogen) contribute to clinical expression of both types of PCT, which suggests that 1 or more of these factors contribute to the generation of a liver-specific inhibitor of URO-D. Compounds that induce specific isoforms of P450 can induce PCT in both humans63 and rodents.60 64 

A liver-specific isozyme of cytochrome P450, P4501A2 (CYP1A2), is essential for chemically-induced porphyria in mice,65 but induction of CYP1A2 alone is not sufficient to produce a porphyric response in rats.66 These data indicate that CYP1A2 is required but not sufficient to produce the rodent porphyric model.

In humans, expression of CYP1A2 is constitutive and highly variable and can be affected by smoking, drugs, cruciform vegetables, and other factors.67 We sought immunohistochemical evidence for increased levels of CYP1A2 in hepatocytes of patients with PCT, but we found no correlation between the level of CYP1A2 and the expression of F-PCT or S-PCT (data not shown). These data, coupled with the findings in the CYP1A2 knockout mouse,65 suggest that some threshold level of CYP1A2 activity is central to the pathogenesis of PCT, but exceeding that threshold has no additional effects.

Both genetic and environmental factors contribute to the pathogenesis of PCT. Two genetic factors have been identified: mutations at theHFE locus and at the URO-D locus. The frequency of homozygosity for the Cys282Tyr HFE mutation is much greater than the frequency of PCT, which indicates that hemochromatosis alone cannot be responsible for the development of PCT. The frequency of URO-D mutations in the population is unknown, but it is unlikely that it is as high as the 29% incidence we found in our PCT population. In pedigrees with F-PCT, most individuals who carry mutantURO-D alleles do not express the porphyric phenotype, which indicates that mutant URO-D alleles alone cannot be responsible for the development of PCT. Unidentified genetic factors probably play a role in the pathogenesis of PCT, as even S-PCT may be familial.9 The proportion of the population exposed to excess alcohol, HCV, and medicinal estrogens is also far higher than the frequency of PCT, indicating that these risk factors alone do not cause PCT. An unclarified interaction between environmental and genetic factors appears to be required for phenotypic expression of PCT in most cases.

Approximately 20% of patients presenting with PCT are homozygous for the Cys282Tyr mutation at the HFE locus, and the majority of PCT patients have been exposed to HCV. These 2 risk factors should be examined in all newly diagnosed patients with PCT. Genotyping at the HFE locus is now widely available through commercial diagnostic laboratories. This method of detecting hemochromatosis homozygotes in association with PCT is preferred, as the iron phenotype is abnormal even in patients with PCT who do not have genetic hemochromatosis (Table 3). Family studies are warranted when a proband with PCT is found to be homozygous for the Cys282Tyr mutation. Iron depletion by repetitive phlebotomy is the preferred therapy for most patients with PCT and may improve the clinical course of HCV infection.