The life expectancy of patients with hemochromatosis is normal if iron is removed while the patient is precirrhotic [4-7], but progress in the ability to make early diagnoses has slowed in the past decade [7]. Further improvement will be possible only with universal screening. We analyzed the frequency of iron overload and iron deficiency in two populations that may be candidates for screening: asymptomatic employees and outpatients in a primary care setting.

Methods

Top Methods Results Discussion Author & Article Info References

Study Participants

We screened 3012 employees (2515 men and 497 women) and 3027 outpatients (1228 men and 1799 women). The employees were from two industrial companies (Mannesmann subsidiaries in Dusseldorf and Muhlheim and Henkel in Dusseldorf) and the Dusseldorf University Hospital. The outpatients were patients of nine primary care physicians (none of the physicians were specialists, such as hepatologists or hematologists). Healthy participants were selected as follows: During a 12-month period, all participants who presented for the initial examination required when entering a company or for a routine checkup required by a company were asked to participate in the study. More than 99% of those invited agreed to participate. For practical reasons, selection in the primary care setting involved only every third patient as patients entered the list at the front desk. Approximately 95% of the invited patients agreed to participate. All participants had to give consent before entering the study, and the study was approved by the local ethical committee. The study protocol excluded persons in whom hemochromatosis or iron deficiency had previously been diagnosed. Participants were asked about previous blood donation; fewer than 3% had donated blood more than once during the previous 5 years, and previous blood donation was not a reason for exclusion.

Laboratory Measurements

Serum ferritin levels were measured by using a commercial enzyme-linked immunosorbent assay (Enzymun, Boehringer Mannheim, Mannheim, Germany). Serum iron levels and total iron-binding capacity were measured by using the Ferrozine method (Boehringer Mannheim) on an automated analyzer (Hitachi 704, Hitachi, Tokyo, Japan).

The upper limits of normal (ferritin level, 250 µg/L for women and 350 µg/L for men; transferrin saturation, 50% for women and 60% for men) were prospectively assessed.

It has been recommended that transferrin saturation values of 50% for women and 60% for men be used as the upper limits of normal [1]. We checked these values by using local data from 251 patients with hemochromatosis and 85 HLA-heterozygous siblings [7]; with this method, 98.4% of homozygotes would have been detected and 94.1% of heterozygotes would have been excluded. In the literature, the upper limits of ferritin levels used for hemochromatosis screening show a wide range: 150 to 250 µg/L for women and 250 to 500 µg/L for men [2, 3]. As upper limits of normal, we used 250 µg/L for women and 350 µg/L for men. We verified these values by using data from 251 patients with hemochromatosis and 85 heterozygous siblings [7]; with this method, 99.6% of homozygotes would have been detected and 94% of heterozygotes would have been excluded.

Both the ferritin level and the transferrin saturation had to be abnormal before further studies, including liver biopsy, were done. We checked the upper limits for the combination of the two values in 251 patients with hemochromatosis and 85 heterozygous siblings [7]; with this method, 98.0% of homozygotes would have been detected and 96.4% of heterozygotes would have been excluded.

All participants with abnormal ferritin levels or transferrin saturation values were asked to return, while fasting, after 4 to 6 weeks of abstinence from alcohol. The rate of false-positive elevations in transferrin saturation is increased after a meal, and ferritin levels are often elevated in persons who abuse alcohol [3]. All participants who had two abnormal values for both ferritin level and transferrin saturation were asked to undergo liver biopsy. The liver iron index was calculated by dividing the liver iron concentration (micromol/g of dry weight) by the patient's age [8]. Presence of cirrhosis and degree of fibrosis were assessed by using accepted methods [9-11]. Participants who refused to undergo liver biopsy received phlebotomy (500 mL of blood/wk) to prove or exclude iron overload. All participants in whom only the ferritin level or only the transferrin saturation was elevated were asked to undergo further evaluation so that we could identify the cause of the laboratory abnormality.

Participants in whom hemochromatosis was proven by liver biopsy or phlebotomy were asked to inform their siblings that they should be evaluated for hemochromatosis by measurement of ferritin levels and transferrin saturation; determination of HLA genotype; and, if necessary, liver biopsy.

Participants with a ferritin level less than 15 µg/L [12, 13] were evaluated for anemia and the cause of iron deficiency.

Statistical Analysis

Multivariable logistic regression analysis was used to investigate the effect of age, sex, and population (patient or employee) on the probability of having an elevated ferritin level or transferrin saturation, respectively, at the initial blood test. For this analysis, age was categorized by quartiles. Statistical analyses were done by using the SPS program (SPS, Inc., Munich, Germany).

Results

Top Methods Results Discussion Author & Article Info References

Screening for Iron Overload

Of the 6039 study participants, 499 (8.3%) had elevated serum ferritin levels, 211 had elevated transferrin saturation, and 55 (0.9%) had both elevated ferritin levels and elevated transferrin saturation at the initial examination. All participants with either abnormality were asked to return, while fasting, after 4 to 6 weeks of abstinence from alcohol.

Evaluation of Participants with Elevated Ferritin Levels and Transferrin Saturation

Figure 1 shows the characteristics of the participants who had increases in both ferritin level and transferrin saturation. Of the 42 participants who had repeatedly abnormal ferritin levels and transferrin saturation values at both tests, 33 had further evaluation. Liver biopsy confirmed iron overload in 17 participants and excluded it in 3 participants (1 of the 3 had chronic hepatitis C, 1 had Niemann-Pick disease, and 1 had chronic granulomatous hepatitis). Thirteen participants did not consent to have liver biopsy and thus received phlebotomy. In 11 of the 13, phlebotomy treatment removed at least 6 g of iron (range, 6.0 to 9.0 g; 500 mL of blood is considered to contain 250 mg of iron), thereby proving iron overload. In the other 2 participants, phlebotomy treatment had to be stopped because anemia occurred after the removal of 1.00 g of iron from 1 participant and 1.75 g of iron from the other. Thus, iron overload was excluded. On these grounds, we identified 28 participants with gross iron overload (Figure 1, Table 1). These 28 participants had 47 living siblings, 32 of whom presented for evaluation. Six of the 32 were HLA identical to the sibling who had hemochromatosis and had abnormal ferritin levels and transferrin saturation values. Further evaluation (liver biopsy in 4 participants and phlebotomy in 2) confirmed hemochromatosis in these 6 siblings.

View larger version (32K): [in this window] [in a new window]   Figure 1. Flow chart of examinations for the detection of iron overload. Percentages given in parentheses are percentages of the values in the previous line, with the exception of the percentages in the third and fourth rows. These values are percentages of the total number of participants.

Only 4 of 21 participants (including the siblings) who underwent liver biopsy had cirrhosis. None of the 13 participants with iron overload who refused to have liver biopsy showed signs of liver cirrhosis on laboratory, sonographic, and clinical examination. Thus, 30 of the 34 participants with gross iron overload were precirrhotic. The 4 cirrhotic participants had various complications of hemochromatosis (Table 1). Of the 30 noncirrhotic participants, 11 had at least one complication due to iron overload (Table 1).

Persons with early-stage hemochromatosis often have not accumulated greatly increased iron stores. According to Edwards and colleagues [1, 3], 60% of screened persons who have abnormal transferrin saturation even in the absence of elevated ferritin levels and hepatic iron concentrations are homozygous for hemochromatosis. In our study, 28 of the 33 participants (84.9%) who underwent liver biopsy or phlebotomy had hemochromatosis and 5 did not. Thus, after we subtract 28 participants identified as iron-loaded and 5 participants with exclusion of iron overload from the 42 participants with elevated ferritin levels and transferrin saturation, 60% of the remaining 9 participants (in a conservative estimate) will be homozygous for hemochromatosis.

Evaluation of Participants with Elevated Transferrin Saturation or Ferritin Levels

At the initial examination, 156 participants had elevated transferrin saturation but normal ferritin levels. Transferrin saturation was again elevated in 92 of the 127 participants who had this value rechecked, and clinical evaluation identified a cause of elevated transferrin saturation in 9 of the 92 (Table 2). In the other 83 participants, annual measurement of transferrin saturation and serum ferritin level was recommended. Of the 444 participants with elevated serum ferritin levels but normal transferrin saturation, 341 presented for a second blood test; this test showed elevated ferritin levels in 136 participants. Further evaluation identified a cause of the elevated ferritin levels in 85 of the 136 (Table 2).

Estimation of the Prevalence of Hemochromatosis

The prevalence of hemochromatosis was estimated separately for male employees, female employees, male outpatients, and female outpatients (Table 3).

Screening for Iron Deficiency

Low serum ferritin levels (<15 µg/L) were found in 208 of the 6039 participants at the initial examination. Data from the further evaluation of these 208 participants are shown in Figure 2.

View larger version (28K): [in this window] [in a new window]   Figure 2. Flow chart of examinations for the detection of iron deficiency and its causes. Percentage values given in parentheses are percentages of the values in the previous line.

Evaluation of Participants with Low Ferritin Levels

Of the 208 participants who had low serum ferritin levels at the initial examination, 178 presented for the second blood test (Figure 2). One hundred sixty-four of the 178 still had decreased ferritin levels. Assuming that all participants would have complied with the check-up and that the number with abnormal ferritin levels on the second test would have been similar to the number among those who presented for the recheck, 192 of the 6039 participants (3.2%) had iron deficiency (1.1% of men and 6.6% of women). The rate of iron deficiency was calculated to be 2.4% in male patients, 6.8% in female patients, 0.5% in male employees, and 6.0% in female employees (Table 4).

Evaluation of the Cause of Iron Deficiency

All 164 participants with repeatedly low serum ferritin levels presented for the suggested clinical evaluation to identify the cause of iron deficiency. The clinical evaluation included medical history, physical examination (including gynecologic examination for women), abdominal ultrasonography, laboratory tests (data not given in detail), and testing for occult blood in the stool.

If the results of all tests were normal and the medical history showed an obvious cause of iron deficiency (Table 2, Figure 2), no further evaluation was done and the participants (n = 95) received oral iron medication and were asked to return at regular intervals. A median of 10 months later, no specific disease was diagnosed in this subgroup. The remaining 69 participants, who did not have an obvious cause of iron deficiency, were evaluated further regardless of whether the results of the clinical examination had been normal or abnormal. Further evaluation included chest radiography, upper gastrointestinal endoscopy, and colonoscopy, in that order. If one of these procedures identified a cause of iron deficiency (Table 2), no further evaluation was done (n = 55). In the remaining 14 participants, however, further evaluation was necessary (not described in detail). A cause of iron deficiency was eventually identified in 10 of these 14 participants. Despite thorough evaluation, the cause of iron deficiency remained unknown in 4 participants (4 of 164 [2.4%]); these participants were treated with oral iron therapy. All 4 had normal ferritin levels after 3 months of iron therapy and were not shown to have a specific disease.

Evaluation of Participants with Anemia Due to Iron Deficiency

Of the 164 participants with iron deficiency, 118 (72.0%) had low hemoglobin levels (20 of 30 men had levels <14 g/dL, and 98 of 134 women had levels <12 g/dL). Assuming that all participants would have complied with the check-up and that the number with abnormal ferritin levels and anemia at the second test would have been similar to the number among those who presented for the recheck, anemia due to iron deficiency was calculated to be present in 2.1% of all participants (1.5% of male patients, 5.1% of female patients, 0.3% of male employees, and 4.0% of female employees) (Table 4).

Relation of Ferritin Level and Transferrin Saturation to Age, Sex, and Population

Elevated ferritin levels were more common in men than in women among both employees (7 of 497 female employees [1.4%] and 257 of 2515 male employees [10.2%]) and patients (107 of 1799 female patients [5.9%] and 128 of 1228 male patients [10.4%]) (adjusted odds ratio, 2.4 [95% CI, 1.9 to 3.1] for men compared with women) (Table 5). Elevated transferrin saturation was more common in female than in male employees (34 of 497 female employees [6.8%] and 38 of 2515 male employees [1.5%]), but rates were similar in male and female patients (88 of 1799 female patients [4.9%] and 51 of 1228 male patients [4.2%]) (adjusted odds ratio, 0.5 [CI, 0.4 to 0.7] for men compared with women). The probability of elevated ferritin levels increased with age independent of sex and population. Age was not associated with risk for elevated transferrin saturation (Table 5).

Discussion

Top Methods Results Discussion Author & Article Info References

Iron deficiency is generally believed to be markedly more common than iron overload. A widely distributed textbook of medicine [14] states that in developed countries, 20% of women and 3% of men have iron deficiency. In previous nutritional surveys [12, 15], iron deficiency was estimated to be present in 5% to 30% of women and 1% to 4% of men. Iron deficiency is reportedly even more common in pregnant women and adolescents of both sexes [12, 14, 16, 17]. Several previous surveys on iron status that were done in blood donors showed that 26% of female persons [18] and 6% of male persons had iron deficiency [13]. In these studies, blood donations probably caused iron deficiency in many participants [18]. In contrast to studies done in blood donors, a recent cross-sectional study done in Icelandic adults showed iron deficiency in only 1% to 2% of men and 2% to 4% of women [19]. It was concluded that Icelandic persons have unusually large stores of iron, due to iron enrichment of flour and cereal, compared with persons in previous cross-sectional studies [12, 20]. Our study shows that the findings in Iceland are not a peculiar phenomenon in a special population but that iron deficiency in developed countries may now be lower than was reported 15 to 20 years ago. The data also show that iron deficiency in adults is usually caused by blood loss, not by dietary factors.

In male employees, gross iron overload was almost as common as iron deficiency (0.4% compared with 0.5%) and was more common than iron-deficient anemia (0.4% compared with 0.3%). In female employees, iron deficiency and iron-deficient anemia were more common than iron overload. The estimated prevalence of homozygous hemochromatosis (including early disease without marked iron overload) in employees was 1% or more for both sexes and thus was, higher than that reported in previous studies of blood donors. The frequency of homozygous hemochromatosis was 1.8 times higher in outpatients with nonspecific symptoms than in employees; this was probably due to selection bias.

Considering the high prevalence of hemochromatosis found in this and other, recent studies, it is important to ask why the disease is not diagnosed more often in clinical practice. Our experience in a large referral center for persons with hemochromatosis indicates that many cases are diagnosed too late or remain undiagnosed. In addition, some persons may have mild disease that remains subclinical. Our data and those from Edwards and Kushner [1] show that the estimated number of homozygotes without relevant iron overload is approximately twice the number of homozygotes who had already accumulated iron to the degree that ferritin levels and hepatic iron concentrations were elevated. Only the follow-up of participants with elevated transferrin saturation but normal ferritin levels will show how many of these participants will develop gross iron overload and clinical disease. It seems likely that some homozygotes never progress to cirrhosis of the liver. Nevertheless, our data show that gross iron overload due to hemochromatosis is also relatively common among employees. Although only 4 of the 34 participants with iron overload had liver cirrhosis, many others already showed some kind of organ damage due to iron overload.

In the past decade, several studies have evaluated the prevalence of hemochromatosis in the general population, which was usually thought to be represented by blood donors [1, 2]. In an analysis of these studies, the mean value for the prevalence of homozygous hemochromatosis was approximately 5.5 per 1000 persons, yielding a gene frequency of 74 in 1000 persons and a frequency of heterozygosity of 134 in 1000 persons [1]. The only study that used serum iron levels as the initial test yielded the lowest prevalence [21]; therefore, we presume that this strategy overlooks many persons with hemochromatosis. Only a few studies have tried to assess the optimal threshold value for screening for hemochromatosis. It has been proposed that a transferrin saturation value greater than 60% for men and greater than 50% for women should be used for the initial test [1, 3]. In that study, liver biopsies were also done in persons who had elevated transferrin saturation but normal ferritin levels. Although liver iron concentrations were normal in almost all of these persons, family studies with likelihood analysis showed that approximately 60% of the latter persons were homozygous for hemochromatosis [1]. Thus, liver biopsy is not useful in persons with abnormal transferrin saturation but normal ferritin levels [1]. Instead, ferritin levels should be checked in these persons annually [3].

We did not formally analyze the cost-effectiveness of screening for hemochromatosis and iron deficiency, but it is evident that such screening will be highly cost-effective even when only the benefit for the 30 participants given a diagnosis of noncirrhotic hemochromatosis is considered. Early diagnosis and iron removal increased life expectancy in this subset of participants by approximately 10 years (yielding 300 years of life saved) [4, 21]. The only relevant cost factor in the screening procedure is the measurement of ferritin levels and transferrin saturation. In the German reimbursement system, these measurements cost approximately $40 in a single person, assuming that no other blood test is done ($40 x 6600 measurements = $264 000). Thus, the cost per year of life saved ($264 000 over 300 years) is less than $1000. This value is favorable compared with those of other medical interventions, such as treatment of hypertension [22, 23]. These considerations on the cost-effectiveness of hemochromatosis screening agree with the results of formal cost-analyses on this topic [21, 24-26].

Whether it is necessary to measure both ferritin levels and transferrin saturation during screening remains to be discussed. It is well documented that transferrin saturation is a more sensitive marker for hemochromatosis than ferritin levels are; ferritin levels become elevated only in the presence of grossly increased body iron stores [1, 3]. Our study corroborates the assumption of Edwards and Kushner [1] that transferrin saturation may be sufficient in screening only for iron overload: Most participants who underwent evaluation because of elevated ferritin levels but who had normal transferrin saturation had diseases other than hemochromatosis [1]. However, several arguments favor the simultaneous measurement of both iron markers. Measuring both ferritin level and transferrin saturation allows for the determination of iron deficiency, which is more common than iron overload in women. In addition, all persons with elevated transferrin saturation have to have ferritin levels measured in the second laboratory test anyway. It was recently reported that some heterozygous carriers of hemochromatosis have elevated transferrin saturation and others have elevated ferritin levels; only a few heterozygotes had both [27]. Finally, in many persons with consistently elevated ferritin levels, a variety of diseases other than hemochromatosis were identified. Thus, it appears more useful to measure both variables instead of measuring only transferrin saturation. Previous studies [1, 3] and our results suggest that it may be advantageous to screen persons while they are fasting because of the potential for postprandial increases in transferrin saturation.

As yet, there is no general agreement about the age at which screening should be initiated. A recent study [28] suggested that screening the children of patients who were homozygous for hemochromatosis is already cost-effective when the children are 10 years of age. On the other hand, iron overload with tissue damage is infrequent in young adolescents [3]. We did not study children or adolescents and thus cannot say whether screening should be done between the ages of 10 and 20 years. Our data show that the frequency of gross iron overload steadily increased with age. The youngest person with iron overload identified in our study was 23 years of age, and another patient was younger than 30 years of age. Thus, screening is already useful at 20 to 25 years of age, although the yield increases at ages greater than 30 years.

Our data indicate that the iron fortification of food products may cause more harm than benefit in men and therefore may not be recommended for the general population. It was recently shown [29] that the interaction of dietary iron and genetic factors may cause severe iron overload. Iron-enriched food products, however, may be useful for premenopausal (menstruating) women, pregnant women, and adolescents.

Our study has some limitations. First, we estimated the frequency of homozygous hemochromatosis by adding the number of firmly diagnosed cases of hemochromatosis and the number calculated by assuming that 60% of participants with elevated transferrin saturation but normal ferritin levels are homozygotes [3]. It is not known whether the latter assumption is exact for this population. Second, this estimation assumed that the frequencies of hemochromatosis and iron deficiency were the same in participants who did not undergo the full evaluation as they were in participants who did. Third, our data cannot be generalized to other populations. A recent study [30] indicates that the frequency of the most prevalent genetic mutations of the HLA-H gene varies widely among different populations. Further studies are needed to determine whether direct genetic tests can be applied in screening programs.

Despite these limitations, our data and those of other recent screening studies strongly suggest that the time has come to establish the use of ferritin levels and transferrin saturation values for the universal screening of hemochromatosis in asymptomatic persons. Screening is at least as cost-effective in primary care outpatients. The broad use of measurements of ferritin levels will also detect iron deficiency, which today is often caused by specific diseases and not by a diet poor in iron.

Dr. Christoph Niederau: Institute for Clinical Chemistry and Laboratory Diagnostics, Heinrich-Heine-Universitat Dusseldorf, Moorenstrasse 5, 40225 Dusseldorf, Germany.

Dr. Lange: Department of Medical Informatics, Biometrics and Epidemiology, Ruhr-Universitat Bochum, 44780 Bochum, Germany.

Dr. Maurer: Betriebsarztzentrum at Mannesmannrohren Werke AG, Rather Kreuzweg 106, 40472 Dusseldorf, Germany.