Effect of a Mild Infection on Serum Ferritin Concentrationð Clinical and Epidemiological Implications

Effect of a mild infection on serum ferritin concentrationÐ clinical and epidemiological implications

L HultheÂn, G Lindstedt, P-A Lundberg and L Hallberg

Department of Clinical Nutrition, Institute of Internal Medicine, and Department of Clinical Chemistry, University of GoÈteborg, Sahlgrenska University Hospital, GoÈteborg, Sweden

Objectives: To study the distribution of serum ferritin concentration in adolescent boys and girls with and without a preceding mild infection. Design: The prevalence of iron de®ciency was studied in two representative samples. The ®rst sample from 1990 comprised 207 boys and 220 girls. The second sample from 1994 included 620 boys and 624 girls. In total 1675 adolescents, 15±16 y old, 827 boys and 844 girls were studied. Results: A signi®cant shift of serum ferritin concentration towards higher values was observed in those who reported an upper respiratory infection with fever during the preceding month (P < 0.001). Signi®cant differences were found between serum ferritin values in healthy, not infected adolescents and serum ferritin values in those with ongoing infection, both in boys and girls in the two materials (P < 0.01), and in those with a mild infection during the preceding three weeks. Conclusions: The prevalence of recent infection should be included as information when trying to assess the prevalence of iron de®ciency on the basis of serum ferritin measurements and when examining relationships between iron status and composition of the diet. The ®ndings imply that differences in prevalence of iron de®ciency between different studies might partly be explained by differences in prevalence of simple respiratory infections. The diagnostic sensitivity of the serum ferritin assay for iron de®ciency, using conventional reference limits, decreases for subjects with recent such infections; similarly, there will be a decrease in the diagnostic speci®city for haemochromatosis. Sponsorship: Swedish Medical Research Council (project B9519X-04721-20B) and Swedish Dairy Association, Stockholm. Descriptors: serum ferritin concentration; mild infection; adolescents; prevalence of iron de®ciency; diagnostic sensitivity and speci®city=iron de®ciency; iron overload

Introduction Some years ago, we made the unexpected observation that even a sore throat combined with fever for a few days The evaluation of iron status requires the use of reliable had a dramatic in¯uence on serum ferritin with an increase indicators. Early studies suggested that the concentration of which remained high for a long time. At our laboratory a serum ferritin was related to the amounts of stored iron nurse on our staff with an iron de®ciency due to heavy (Walters et al, 1973). Formulas have been devised to menstrual bleedings participated in a study to reduce the estimate iron stores from concentrations of serum ferritin bleedings. Her serum ferritin was followed monthly for in healthy individuals (Cook & Finch, 1979; Cook et al, three months and was in the range 10±12 mg/L. One week 1986). The introduction of serum ferritin for the evaluation after the start of her infection combined with fever for some of the iron status (Addison et al, 1972; Jacobs et al, 1972) days, her serum ferritin increased to 75 mg/L. At that time was a marked diagnostic improvement since low serum she was afebrile, clinically well and back to work. There- ferritin concentrations are only seen in iron de®ciency, after serum ferritin was followed weekly and remained indicating an absence of iron stores. Low serum ferritin is elevated for eight weeks. Serum ferritin then turned back to not associated with any other known condition (Walters et the initially low value, 11 mg/L. This observation suggested al, 1973). On the other hand, it is well-known that, for that even a rather mild infection with fever may markedly example, severe infections, in¯ammatory conditions, liver increase serum ferritin and may remain elevated for several diseases, starvation and heavy ethanol intake, could weeks. In our recent studies on adolescents we used serum increase serum ferritin also in iron de®cient subjects, ferritin as a main marker of iron status. Our observation (BirgegaÊrd et al, 1978; Worwood, 1979; Lundberg et al, above suggested that even a mild infection might in¯uence 1984; Lundin et al, 1981; Meyer et al, 1984). the level of serum ferritin. An effect of mild infections on serum ferritin, however, The purpose of this present study was to examine the has not been examined. Such an effect on the prevalence of effect of a preceding mild infection on serum ferritin iron de®ciency in population studies as well as on its values. This was made both by comparing the distribution diagnostic sensitivity and speci®city for abnormal iron of serum ferritin in those with and without a preceding stores, has not been examined. infection in the last month and by examining the difference in serum ferritin as above but in relation to the interval Correspondence: Dr L HultheÂn. Received 20 December 1997; revised 4 February 1998; accepted between the examination of serum ferritin and the time of 28 February 1998 the infection. Effect of a mild infection on serum ferritin concentration L HultheÂn et al 377 Material and methods De®nition of infection was based on an examination and on a medical history, taken by a specially trained nurse, the Two samples of 15 to 16 y old boys and girls in GoÈteborg same for both the studies. In the questionnaire used, were studied. The ®rst study was performed in 1990 in questions were raised about the present state of health early spring when the incidence of infections is usually and about recent disorders. A speci®c question was raised low. This study comprised four schools and the sample about recent infection and if so, the time for the infection represented 1 out of 10 of all pupils in the 9th grade that was registered as well as its severity (fever for a certain year in GoÈteborg. Written information was sent via all the number of days, symptoms and treatment). pupils to their parents, who had to give written approval for drawing of blood samples. Permission was granted for 220 girls (86% of those invited in the original sample) and 207 Statistics All statistical analyses were made using a Stat- boys (80% of those invited in the original sample). view IV computer program (Abacus Concepts Inc., Ber- The ®rst study was a pilot study, comprising different kely, CA). Mean values were compared using the t-test. socioeconomic groups. Due to its size, moreover, the The project was approved by the Ethical Committee of sample was less representative of the general population. the Medical Faculty of the University of GoÈteborg. In the second study greater efforts were made, in cooperation with the epidemiological unit of the town council, to get a representative sample, covering different Results socioeconomic and living conditions. The second study was In the ®rst study, made during March±April 1990, 21.7% of performed in October±December 1994, and included 624 the 207 boys and 20.5% of the 220 girls reported an upper girls (90% of those invited to the study) and 620 boys (89% repiratory infection with fever during the preceding month. of those invited). This second study was more comprehen- The infection rate was signifcantly higher among those sive, and comprised 13 schools chosen to represent areas with a high serum ferritin (P < 0.001), suggesting that a with different socioeconomic and living conditions. All preceding infection may have shifted serum ferritin toward adolescents in grade nine at these schools were invited to higher values. Among the girls with no history of infection participate. The participants represented 24.8% of the total (N 175) 43.9% had serum ferritin, 15 mgL vs 25.5% number of adolescent students in this age group in GoÈte- among those with a history of infection (N 45). For the borg that year. GoÈteborg is the second largest city in boys with no history of infection (N 162) 17.0% had Sweden with about 430 000 inhabitants. serum ferritin 15 mg/L vs 4.8% among those with a This study was carried out to get knowledge about the history of infection (N 45). baseline iron status situation in adolescents in Sweden In the second study, performed during October±Decem- before the general iron forti®cation of white wheat ¯our ber 1994, 19.8% of the 620 boys and 21.8% of the 627 girls was stopped (31 December 1994). To study the effect of reported an upper respiratory infection during the preceding withdrawal a new study is planned for 1998. month. We found a marked shift of serum ferritin towards higher values in those who reported an upper respiratory Serum ferritin used in the two studies was determined by a infection during the preceding month (P < 0.001). Among double-antibody polyethylenglycol RIA (Diagnostic Pro- the girls with no history of infection (N 501) we found ducts Corp. Los Angeles, CA). The assay was calibrated 39.6% with serum ferritin 15 mg/L vs 16% among those against World Health Organization 1st International Stan- with a history of infection (N 140). For the boys with no dard, IS80/602 (Worwood, 1979). history of infection (N 491) we found 24.7% with serum

Table 1 Prevalence of iron de®ciency, serum ferritin 15 mg=L. With and without reported preceding mild infection

Without preceding infection*** With preceding infection*** The total material

(no) (%) (no) (%) (no) (%)

Boys* 1990 28=165 17.0 2=42 4.8 3 0=207 14.5 Boys* 1994 121=489 24.7 1 0=131 7.6 13 1=620 21.1 Girls** 1990 76=173 43.9 1 2=47 25.5 8 8=220 40.0 Girls** 1994 197=498 39.6 2 7=125 16 22 4=623 35.9

*P < 0.05 (1990 vs 1994). **ns (1990 vs 1994). ***P < 0.001.

Table 2 Serum ferritin in adolescents from the 1990 study, grouped according to infection

Serum ferritin (boys) Serum ferritin (girls)

mean s.d. n P* mean s.d. n P*

Healthy, not infected 21 7.8 162 15 9.5 175 Ongoing infection 30 18.0 6 < 0.01 20 3.6 4 < 0.01 Infection 1 week ago 36 11.6 22 < 0.001 39 20.8 25 < 0.001 Infection 2 weeks ago 46 21.4 9 < 0.001 19 9.8 7 < 0.05 Infection 3 weeks ago 54 7.9 3 < 0.001 24 10.5 6 < 0.05 Infection 4 weeks ago 22 6.8 5 ns 16 6.7 3 ns

*P-value: not infected vs infected. Effect of a mild infection on serum ferritin concentration L HultheÂn et al 378 In both samples serum ferritin values in the group with ongoing infection were signi®cantly higher compared with the healthy, non-infected group. The same was found for the groups with a history of infection one week ago, two weeks ago and three weeks ago, with exception of the boys in the 1994 sample. There was found no statistically signi®cant difference in serum ferritin values in those with a history of infection four weeks ago and the healthy, not infected group (Tables 2 and 3). Discussion Studies on the prevalence of iron de®ciency were pre- viously based mainly on hemoglobin determinations and focused on the prevalence of anemia. The balance of evidence indicates that the main negative effects of iron de®ciency may not be related to a reduced supply of oxygen to tissues by a reduced concentration of hemoglo- Figure 1 Distribution of mean decile values of log serum ferritin bin in the blood but probably rather to an insuf®cient concentration in 207 boys and 220 girls with or without a history of a content of iron in tissues especially iron containing or preceding mild infection (1990 study). iron dependent enzymes. There are recent well-controlled studies supporting this interpretation (Ballin et al, 1992; Rowland et al, 1994; Zhu & Haas, 1997). Iron de®ciency is de®ned as absence of iron stores, combined with signs of an iron de®cient erythropoiesis indicating an insuf®cient supply of iron to all tissues including the erythron (Hallberg et al, 1993a). In an extensively studied random sample of women, all aged 38 y, a single criterion of serum ferritin 15 mg/L was found to best discriminate between iron de®cient and iron replete women, based on absence/presence of stainable iron in bone marrow smears (Hallberg et al, 1993a). The same criterion was used in the present studies in teenagers, since the relationship between serum ferritin and transferrin saturation was the same in teenagers as in adult women (Hallberg et al, 1993b). Moreover, by relating Hb, MCH and MCV to serum ferritin we found signs of an iron de®cient erythropoiesis, appearing at the same serum ferritin (Hallberg et al, 1993b). These facts suggest that the same cutoff value for serum ferritin should be used in Figure 2 Distribution of mean decile values of log serum ferritin teenagers as in adults. concentration in 620 boys and 623 girls with or without a history of a The origin of circulating ferritin is still unknown, as is preceding mild infection (1994 study). (are) the route(s) for its elimination from the plasma compartment and factors affecting its elimination rate. For instance, the increased concentrations observed in ferritin 15 mg/L vs 7.6% among those with a history of individuals with a reduced caloric intake (Lindstedt et al, infection (N 123). The differences observed were statis- 1980; Lundberg et al, 1984) might well be due to slowed tically signi®cant (P < 0.001) (Table 1). rate of elimination. The balance of evidence strongly In both studies and for both boys and girls, the serum suggests that the concentration of serum ferritin in normally ferritin distribution curves were shifted toward higher related to properties of a labile iron pool which in turn is values in those with a history of an infection (Figure 1, in¯uenced by the main iron stores. The reticulo-endothelial Table 1 and Table 2). We also examined the effect of the system may be the main compartment of the labile iron interval between the time of the infection and the sampling pool, which may explain for example, why infections and of the blood to analyse serum ferritin. This was valid both in¯ammations in the body markedly in¯uence serum ferri- for the 1990 and the 1994 samples (Table 2 and Table 3). tin.

Table 3 Serum ferritin in adolescents from the 1994 study, grouped according to infection

Serum ferritin (boys) Serum ferritin (girls)

mean s.d. n P* mean s.d. n P*

Healthy, not infected 27 17.2 491 22 13.8 501 Ongoing infection 36 20.6 54 < 0.01 30 21.2 57 < 0.01 Infection 1 week ago 49 21.3 35 < 0.01 36 19.6 27 < 0.001 Infection 2 weeks ago 39 17.9 16 < 0.05 35 18.9 31 < 0.001 Infection 3 weeks ago 41 24.5 10 ns 35 12.2 13 < 0.001 Infection 4 weeks ago 31 12.3 8 ns 23 16.0 12 ns

*P-value: not infected vs infected. Effect of a mild infection on serum ferritin concentration L HultheÂn et al 379 Conclusions BirgegaÊrd G, HaÈllgren R, Killander A, StroÈmberg A, Venge P & Wide L (1978): Serum ferritin during infection. A longitudinal studie. Scand. J. Using serum ferritin in the diagnosis of iron de®ciency in Haematol. 21, 333±340. clinical practice, present ®ndings show that even mild Cook JD & Finch CA (1979): Assessing iron status of a population. Am. J. infections should be taken into account when interpreting Clin. Nutr. 32, 2115±2119. Cook JD, Skikne BS, Lynch SR & Reusser ME (1986): Estimates of iron the data. A subnormal serum ferritin, however, in close suf®ciency in the US population. Blood 68, 726±731. connection to an infection is of course still a sign of iron Hallberg L, HultheÂn L, Lindstedt G, Lundberg P-A, Mark A, Purens J, de®ciency but a normal serum ferritin does not exclude an Svanberg B & Swolin B (1993a): Prevalence of iron de®ciency in iron de®ciency. When assessing iron status of populations Swedish adolescents. Pediatri. Res. 34, 680±687. Hallberg L, Bengtsson C, Lapidus L, Lindstedt G, Lundberg P-A in epidemiological studies, present results indicate that & HultheÁn L (1993b): Screening for iron de®ciency: an analysis even mild infections with fever during the preceding based on bone-marrow examinations and serum ferritin determi- three weeks must be considered. n-ations in a population sample of women. Br. J. Haematol. 85, 787± Differences in prevalence of iron de®ciency between 798. different studies might actually be explained by differences Jacobs A, Miller F, Worwood M, Beamish MR & Wardrop CA (1972): Ferritin in the serum of the normal subjects and patients with iron in prevalence of simple respiratory infections such as a de®ciency and iron overload. Br. Med. J. 4, 206±208. common cold for just a couple of days. Lindstedt G, Lundberg P-A, BjoÈrn-Rasmussen E & Magnusson B (1980): Our ®ndings also suggest that in studies on causes and Serum-ferritin and iron-de®ciency anaemia in hospital patients. Lancet effects of iron de®ciency the prevalence of recent infection i, 205±206. Lundberg P-A, Lindstedt G, Andersson T, BranegaÊrd B & Lundquister G must be taken into account. A proper classi®cation of iron (1984): Increase in serum ferritin concentration induced by fasting. de®cient and non-iron de®cient subjects is important to Clin. Chem. 30, 161±163. determine the true prevalence of iron de®ciency and to Lundin L, Hallgren R, BirgegaÊrd G & Wide L (1981): Serum ferritin in examine relationships between iron status and composition alcoholics and the relation to liver damage, iron state and erythropoietic of the diet, such as intake of iron or foods/nutrients known activity. Acta. Med. Scand. 209, 327±331. Meyer TE, Kassianides C, Bothwell TH & Green A (1984): Effects of to enhance or inhibit iron absorption, for example ascorbic heavy alcohol consumption on serum ferritin concentrations. A. Afr. acid, meat, calcium and phytate. A misclassi®cation may Med. J. 66, 573±575. also mask results in analysis of differences in ®ndings and Rowland TW, Deisroth MB, Green GM, Kelleher JF (1998): The effect of symptoms in normal and iron de®cient subjects. ion therapy on the exercise capacity of non anemic iron-de®cient adolescent runners. Am. J. Dis. Child 142, 165±169. Walters GO, Miller FM & Worwood M (1973): Serum ferritin References concentration and iron stores in normal subjects. J. Clin. Pathol. 26, 770±772. Addison GM, Beamish MR, Hales CN, Hodgkins M, Jacobs A & Worwood M (1979): Serum ferritin. Crit. Rev. Clin. Lab. Sci. 10, Llewellin P (1972): An immunonoradiometric assay for ferritin in the 171±204. serum of normal subjects and patients with iron de®ciency and iron Zhu YI & Haas JD (1997): Iron depletion without anaemia and physical overload. J. Clin. Pathol. 25, 326±329. performance in young women. Am. J. Clin. Nutr. 66, 334±341. Ballin A, Berar M, Rubenstein U, Kleter Y, Hershkovitz A & Meytes D (1992): Iron state in female adolescents. Am. J. Dis. Child 146, 803±805