Hyperproduction of erythropoietin in nonanemic lead-exposed children.

Lead (Pb) poisoning has numerous effects on the erythropoietic system, but the precise mechanism whereby high dose exposure causes anemia is not entirely clear. We previously reported that Pb exposure is associated with depressed serum erythropoietin (EPO) in pregnant women residing in a Pb mining town and in a nonexposed town in Kosovo, Yugoslavia. In a prospective study, we tested the hypothesis that blood Pb concentration (BPb) may be associated with depressed EPO in children. BPb, hemoglobin (Hgb), and serum EPO were measured at ages 4.5, 6.5, and 9.5 years in 211, 178, and 234 children, respectively. At 4.5 years of age, mean BPbs were 38.9 and 9.0 microg/dl in the exposed and nonexposed towns, respectively; BPbs gradually declined to 28.2 and 6.5 microg/dl, respectively, by age 9.5 years. No differences were found in Hgb at any age. At age 4. 5 years, a positive association between BPb and EPO (beta = 0.21; p = 0.0001), controlled for Hgb, was found. The magnitude of this association declined to 0.11 at age 6.5 years (p = 0.0103) and 0.03 at age 9.5 years (p = 0.39). These results were confirmed using repeated measures analyses. We concluded that in Pb-exposed children, the maintenance of normal Hgb requires hyperproduction of EPO. With advancing age (and continuing exposure), this compensatory mechanism appears to be failing, suggesting a gradual loss of renal endocrine function due to Pb exposure.

function due to Pb exposure. Key work erythropoietin, hematopoiesis, lead, red cel survival.
Several studies suggest important subclinical, perhaps preclinical, defects in hematopoiesis. Grandjean et al. (11) described delayed blood regeneration capacity in Pb-workers who had normal Hgb concentrations prior to blood donation. That report led us to hypothesize that Pb may inhibit the synthesis of erythropoietin (EPO), a glycoprotein hormone which regulates both steady-state and accelerated erythrocyte production. More than 90% of EPO is produced in the proximal renal tubule (12,13) where Pb accumulates; the remainder is produced in the liver. Indeed, we previously demonstrated that EPO is significandy depressed among pregnant women with moderately elevated BPbs (14).
Hu et al. (15) subsequently described a highly significant negative association between tibia (and patella) bone Pb concentration and Hgb among 119 nonanemic men from the Carpenters' Union; yet no association between BPb and Hgb was found. This report suggests that bone Pb may be a more important marker of ongoing Pb toxicity than BPb. The authors suggested that the observed associations may reflect inhibition of hematopoiesis through depression of EPO, with bone Pb serving as a proxy for kidney Pb (15).
As part of a long-term prospective study of environmental Pb exposure during infancy and childhood (16)(17)(18), we examined the associations between BPb and EPO at ages 4.5, 6.5, and 9.5 years. We found that children with elevated BPbs maintain normal Hgb, but require hyperproduction of EPO to do so. Our working hypothesis is that children with moderately elevated BPbs have shortened red cell survival, a phenomenon previously described only in children with Pb encephalopathy (9) and in Pb workers (10).

Methods
This study was conducted in two towns in Kosovo, Yugoslavia: Kosovska Mitrovica (K. Mitrovica), the site of a Pb mine, smelter, refinery, and battery plant; and Pristina, a relatively nonexposed town 25 miles to the south. Children were selected for follow-up from a previous prospective study of 1,502 pregnant women residing in these towns. Pregnancy outcomes (19,20) and childhood developmental outcomes have been previously described (16)(17)(18).
Subjects. In brief, 706 mother-infant pairs from the pregnancy study were invited to participate in a follow-up study involving repeated visits at 6-month intervals. Of these, the parents of 541 consented and brought their children to at least one visit. Of those who consented, 311 (53.5%), 267 (49.4%), and 260 (48.1%) participated in the visits at ages 4.5, 6.5, and 9.5 years, respectively. For the present study, whole blood for BPb and Hgb was available for 272, 201, and 234 children at each of these ages, respectively; sera for EPO analyses were available for 211, 178, and 234 children, respectively.
Laboratory analyses. At mid-pregnancy, delivery, and at each 6-month follow-up visit, venous blood samples were taken for the measurement of BPb (21), erythrocyte protoporphyrin [EP (24], and Hgb. At each visit, additional sera were obtained and frozen immediately to create a serum bank, from which the current study is derived. Whole blood and serum samples were stored at 40C and -20°C, respectively, and appropriately transported to Columbia University where all assays were performed. The laboratory participates in the BPb and EP quality control program of the Centers for Disease Control. During the study period relevant to this analysis, agreements with the quality control values for BPb and EP, measured by intradass correlation coefficients, were both 0.99. All available sera from the visits at ages 4.5, 6.5, and 9.5 years were analyzed in duplicate for EPO using a commercially available enzyme immunoassay (23). In our laboratory, the limit of detection of the assay was 0.6 mIU/ml, and the coefficient of variation for duplicate measures was 9.8%, 8.3%, and 7.9%, respectively, at ages 4.5, 6.5, and 9.5 years.
Statistical analysis. The association between BPb and EPO was examined using regression methods. Because the distributions of EPO and BPb are skewed, logarithmic transformations were used for these variables. All analyses were performed using SAS version 6.12 (SAS Institute, Cary, NC). First, we evaluated the association between BPb and EPO at each age using linear regression analysis, controlling for concurrent Hgb concentration. Concurrent Hgb was always controlled because it is the most important predictor of EPO (13). Second, we combined data from all ages and used repeated measures analysis (described below) to determine whether the associations between BPb and EPO, controlling for Hgb, changed over time.
The repeated measures analysis was based on 614 observations from 346 children with EPO, BPb, and Hgb measurements available for at least one time point (i.e., ages 4.5, 6.5, or 9.5 years). The 614 observations induded 157 (45.5%) children who had measurements made at one time point (i.e., one age), 110  measures analysis was performed using the marginal linear model [GEE (24)], which takes into account the within-subject correlations between these variables; the details of this analysis and its findings are provided in the Appendix. For the 15 subjects with missing values for BPb or Hgb at a specific time point, we used the following algorithm to substitute values. If BPb (or Hgb) measurements were available for both the 6 months prior to and the 6 months after the missing values, we substituted the mean of these measurements. If BPb (or Hgb) was available for only one of these time points, that value was substituted. Thus, substitutions were made for 12 of these 15 subjects.

Results
During the course of the 5-year period, mean BPb in each town dedined by approximately 28% (Fig. 1A). Mean Hgb in each town remained constant and within the normal range (Fig. 1B). At ages 4.5, 6.5, and 9.5, BPbs ranged from 4.6 to 73.1, 3. Error bars indicate standard error. to 14.5, 10.7 to 15.1, and 9.6 to 15.3 g/dl, respectively; and EPO concentrations ranged from 2.4 to 40.4, 1.6 to 27.1, and 2.1 to 24.0 mU/ml, respectively. In Pristina, the mean EPO rose steadily over time from 5.6 to 8.8 mIU/ml (Fig. IC). In K. Mitrovica, EPO rose between ages 4.6 and 6.5 years (from 7.8 to 9.7 mIU/ml) and then declined to 8.8 mIU/ml at age 9.5 years. The differences in EPO between towns were statistically significant at ages 4.5 (p = 0.0017) and 6.5 years (p = 0.02), with higher concentrations in K. Mitrovica.
Because Hgb is a strong determinant of EPO (13), EPO was examined within Hgb strata for each age (Fig. 2). Few children were anemic; there were 9, 1, and 1 children with Hgb below 11.0 g/dl at ages 4.5, 6.5, and 9.5 years, respectively. Within each Hgb stratum, EPO was higher in K. Mitrovica except at age 9.5 years, when the town differences diminished.
Associations between BPb and EPO, adjusted for Hgb and specific to each age, are shown in Table 1. As expected in each of these models, Hgb was inversely and significandy associated with EPO. Statistically significant associations between BPb and EPO were found at ages 4.5 and 6.5 years; however, the magnitude of the association at age 6.5 was approximately halfthat at age 4.5. At age 9.5 years, the association was not statistically significant and was approximately an order of magnitude less than that at age 4.5.
Results of a regression analysis using repeated measures were similar. At ages 4.5 and 6.5 years, BPb was significantly and positively associated with EPO (p<0.0001 and p = 0.0007, respectively). This association diminished at 9.5 years of age. Indeed, the difference in the association between BPb and EPO at ages 4.5 and 9.5 years was statistically significant (p = 0.0006), while that between ages 4.5 and 6.5 was not. To test whether the association between BPb and EPO varied by Hgb, we repeated all analyses indcluding the appropriate interaction terms.
Based on this GEE model and after adjustment for Hgb, a significant interaction between BPb and age was found, indicating that the association between EPO and BPb varied by age (see Appendix). Further analyses found no interaction ofHgb and age.

Discussion
We previously reported that Pb exposure was associated with depressed serum EPO in pregnant women (14). Our follow-up prospective study of their children provided a unique opportunity to longitudinally examine subtle hematologic effects of Pb exposure in a population whose lifetime exposure has been well characterized. In the current analyses, we observed that serum EPO, after adjustment for Hgb, was positively associated with BPb in children at ages 4.5 and 6.5 years. The findings imply that in nonanemic Pb-exposed children, increased erythrocyte production is required to maintain normal Hgb concentrations. There is strong evidence, discussed below, to support the condusion that this compensatory mechanism is a response to Pb-induced shortened red cell survival. With age, however, the adjusted regression coefficients between BPb and EPO (Table 1) declined from 0.21 at 4.5 years (p = 0.0001) to 0.11 at 6.5 years (p = 0.0103), to 0.03 at age 9.5 years (p = 0.39). It is possible that this gradual decline in slope was due to the modest decline in mean BPb as children aged, i.e., that BPbs fell below a critical threshold for the putative Pb effect. However, our previous study in pregnant women (14) found depressed serum EPO in women whose BPbs were substantially lower than the children in the current study. Thus, it appears more likely that the dedine in slope with age reflects a gradually decreasing ability to produce EPO. We also note, however, that the relationship between BPb and EPO is subtle because the observed serum EPO concentrations in this cohort were nearly all within the published age-related norms (25). A review by Aub et al. in 1925 (1) concluded that although severe plumbism in man may involve "bone marrow failure," anemia was initially due to accelerated destruction of circulating red cells, a phenomenon first suggested in the 19th century (26,27). Supportive mechanistic studies revealed that patients with plumbism had red cell membrane defects which were associated with increased red cell mechanical fragility and decreased osmotic fragility (1). In animals, Pb-induced anemia was found to be reversed by removal of the spleen (28), which removes abnormal and/or aged red cells from circulation.
Numerous animal studies indicate that a two-step sequence of events is involved in Pb-induced anemia. For example, in rabbits acutely poisoned with Pb, peripheral red cell destruction was followed by active bone marrow hyperplasia (29), indicative (in retrospect) of active renal EPO production. With time, however, rabbit bone marrow activity declined, suggesting either failure of erythropoietic stem cells, a decline in renal EPO production, or both. The same temporal sequence has been described in children. Leikin and Eng (9) described shortened 51Cr red cell survival in children with BPbs of 60-238 pg/dl. Among their patients whose Pb exposure was short in duration, turnover of 59Fe in plasma was abnormally high, indicative ofbone marrow hyperactivity; in those with long-term Pb exposure iron turnover was depressed, consistent with marrow hypoactivity.
Our finding that BPb was positively associated with EPO in children at ages 4.5 and 6.5 years is consistent with the first pathophysiologic step, i.e., peripheral red cell destruction, renal EPO production, and bone marrow hyperactivity. Our earlier finding that EPO is negatively associated with BPb in moderately anemic pregnant women [average age = 26 years (14) .00 The results.of this ansi ar costn wit th ciml regression mnodels presented in the main body of the paper. EPO ap'pears to increase with increas ing BPb at ages U. and 15 years, .but.that relationship flattens out at age 9.5 years.. Moreover,the regression .coefficient at aegs 4.5 and 9.5 years are significa M iferent sugstn st diffrec in the effctof B.Pb ths gs step, i.e., bone marrow hypoactivity due to chronic renal, and possibly marrow, Pb toxicity. We speculate that the lack of association between BPb and EPO at age 9. This position offers an excellent opportunity to build upon and implement a commitment to education and research shared by the faculty and university administration. Implementation of this vision will be supported by the commitment of new funding and the potential to recruit new faculty.
Leading candidates will have distinguished records of accomplishment in higher education and/or other fields of professional endeavor and will have demonstrated significant leadership skills. Qualifications for the position include an earned doctorate in Environmental Health or related disciplines; a record of teaching, service, research and scholarly activities; individual and collaborative research; and administrative experience.