Background

Renal failure is an important health concern for persons with hypertension and diabetes. In the fawn-hooded rat, a renal failure locus, Rf-1, has been identified on rat chromosome 1. A study of African American sibpairs with end-stage renal disease (ESRD) replicated this finding on the orthologous region in humans, chromosome 10, with a maximum logarithm of odds (LOD) score of 3.4. An important question is whether this region can be detected in healthy subjects prior to onset of ESRD by examining creatinine clearance as an indicator of early renal damage.

Methods

We analyzed 49 Utah Caucasian pedigrees and performed quantitative nonparametric linkage analysis using 21 markers spanning chromosome 10. Pedigree members (mean age of 40 17) were examined up to three different times over 10 years, with creatinine clearance measured at each exam. For examination 1, three overnight, timed, 12-hour urine samples were obtained and averaged. One 12-hour sample was obtained for examinations 2 and 3.

Results

Heritabilities of creatinine clearance were 0.33 (N = 1360), 0.36 (N = 1196), and 0.53 (N = 718) for the three examinations, respectively. The nonparametric LOD score for examination 1 was 1.4 at marker D10S677 (117 cM). The LOD score at examination 2, an average of 2¹/₂ years later, was 1.8 at marker D10S1239 (123 cM) and 1.9 at marker D10S1425 (137 cM). The LOD score at examination 3, an average of 10 years from baseline, was 2.1 at marker D10S2470 (113 cM). Thus, there is consistent evidence of linkage to this region from three different examinations spanning a period of 10 years.

Conclusions

These linkage results confirm the ESRD linkage and the rat renal failure linkage to this region even though the LOD score is somewhat weaker, probably due to the less severe phenotype that was analyzed. It also suggests that there may be a locus on chromosome 10 that leads to reduced renal function that can be detected while subjects are still healthy. Identification of the responsible gene may help in predicting renal disease progression in susceptible patients.

METHODS

Beginning in 1980, 2500 members of 98 Utah pedigrees were examined at the Cardiovascular Genetics Clinic of the University of Utah. Most pedigrees were ascertained for two or more early coronary heart disease deaths or two or more stroke deaths in the sibship of the founding generation of the pedigree. A few pedigrees were ascertained for the presence of hypertension in a proband⁶. In 1983, over 90% of these persons were re-examined in our clinic. Another examination of a subset of these subjects began in 1990 for a 10-year follow-up of disease end points and changes in clinically measured variables. Since the major goal of this 10-year follow-up was to obtain hypertension incidence data, older family members were targeted for clinical examination, although the vital status of 99% of the original cohort was updated. Almost 33% of the original cohort of 2500 family members were under the age 18 at the first examination and were only recruited for the third examination if they lived close to the clinic. Only 60% of examination 1 subjects were examined at the third examination. Descriptions of additional variables measured and characteristics of the ascertained pedigrees may be found in other references^6,7,8.

Serum creatinine was measured in the morning after an overnight fast at the clinical laboratory of the University of Utah hospital. At examination 1, timed, 12-hour, overnight urine samples were collected from three different days. Two samples were collected from weekdays and one sample from the weekend. Urine volume and creatinine were measured from each sample, with the resulting measurements averaged and converted to amount of creatinine per 24 hours. Creatinine clearance was measured as the ratio of 24-hour urine to serum creatinine amounts. Examinations 2 and 3 collected only a single 12-hour overnight sample. Laboratory autoanalyzers changed between examinations 2 and 3. At examination 3 the Cardiovascular Genetics Laboratory measured serum and urine creatinine on a Roche Fara autoanalyzer and no machine comparisons were done at the time (1991). Therefore, the creatinine clearance at examination 3 should not be directly compared to examinations 1 and 2, but within examination measurements are consistent for all subjects. Longitudinal changes in creatinine clearance were calculated as simple differences of examination 3 minus examination 1 or examination 2 minus examination 1 measurements.

Creatinine clearance measurements and differences between examinations were adjusted for gender, baseline age, age², age³ and body mass index (BMI) by multiple regression analysis. The unadjusted mean was added to the residuals. Excluded from analysis in this study were all subjects under the age of 18 at examination 1 and persons with diabetes (N = 21) at examination 1. Only three subjects became diabetic between examinations 1 and 2 and three more between examinations 2 and 3 and these persons were retained in the analysis. Persons with a baseline history of recurrent urinary tract infections or any form of diagnosed kidney disease were also excluded (N = 79). Two subjects with very high creatinine clearance were excluded because they appeared to be outliers in the distribution. No adjustment was made for hypertension, which was present in 7%, 9%, and 14% of subjects at examinations 1 to 3, respectively.

The Mammalian Genotyping Service⁹ provided 714,335 marker genotypes on 421 markers, 1855 individuals, and 49 pedigrees after data cleaning. Twenty-one markers were analyzed on chromosome 10. Details of the markers and genetic map can be obtained from the Mammalian Genotyping Service website (http://research.marshfieldclinic.org/genetics). Pedigree relationships were assessed by the ASPEX program (ftp://lahmed.stanford.edu/pub/aspex/index.html) and a modified version of PAP¹⁰ and marker compatibilities were checked by PEDCHECK¹¹. There were 1360 persons with both marker and creatinine clearance data from examination 1, 1196 from examination 2, and 718 from examination 3. Multipoint haplotypes were estimated by the Markov chain Monte Carlo methods of Göring and Terwilliger¹² and Thomas et al¹³ to help differentiate identity-by-descent from identity-by-state sharing at each marker. A nonparametric statistic of linkage was calculated at each marker using these haplotypes following the method of Kruglyak et al¹⁴, but extended for quantitative traits by Camp et al¹⁵ using the following formula for each pedigree and marker position:

The mean creatinine clearance of the pedigrees used for q₀, q_i is the quantitative trait value, the first sum is taken over all haplotypes (h) in the pedigree, and the second sum is taken over individuals (i) carrying haplotype (h).

RESULTS

Table 1 shows the clinical characteristics of the sample at examination 1. Serum creatinine increased slightly over the three examination periods. As mentioned above, direct comparisons of urine creatinine cannot be made between examinations 1 or 2 and examination 3 because of laboratory assay changes over the 10-year period. The gender distribution at examination 1 was 51% female and 49% male. All subjects were Caucasian. There were only three males with a serum creatinine greater than 1.6 mg/dL [one at examination 2 (1.9 mg/dL) and two at examination 3 (1.7 and 2.2 mg/dL)]. Only one female had a serum creatinine greater than 1.4 mg/dL (1.7 mg/dL at examination 3). Gender, a cubic polynomial in age, and BMI explained 43%, 24%, and 9% of the creatinine clearance variance at examinations 1, 2, and 3, respectively.

Table 1 - Characteristics (mean SD) of the subjects used for the linkage analysis of creatinine clearance.

Full table

Comparisons of creatinine clearance among individuals from each of the three methods of pedigree ascertainment showed that the coronary heart disease- and hypertension-ascertained pedigrees had significantly lower baseline means ( SE) than stroke-ascertained pedigree members (128 1.5, 132 2.0, and 139 2.2 mL/min, respectively). There were no differences among pedigree ascertainment methods for longitudinal changes in creatinine clearance.

Table 2 shows the significant heritability estimates of creatinine clearance for each of the three examinations, as well as for the changes between examinations. The heritability estimates for the longitudinal changes were significant despite the correlations of creatinine clearance between examinations 1 and 2 and examinations 1 and 3 being only 0.20 and 0.18, respectively. If heritabilities are calculated using only subjects who had all three examinations so that the heritabilities among examinations were more directly comparable, the estimates were 0.32, 0.27 and 0.56 for examinations 1, 2, and 3, respectively.

Table 2 - Heritabilities ( SE) of creatinine clearance and changes in creatinine clearance for the three examinations.

Full table

Figure 1 shows the linkage plots for creatinine clearance for each of the three examinations. Examination 3 shows the best signal for linkage (2.09) at marker D10S2470 located at 113 cM on the Marshfield map. Examination 2 showed two peaks further downstream (1.82 and 1.94) but clearly overlapping the examination 1 signal. The maximum examination 1 nonparametric LOD score (1.42) occurred at marker D10S677 at 117 cM, the marker next to the examination 3 peak.

Figure 1.

Nonparametric logarithm of odds (NPL LOD) score plots of chromosome 10 markers and creatinine clearance measurements from each of the three clinical examinations. Symbols are: (), examination 1; (), examination 2; (), examination 3.

Full figure and legend (51K)

Figure 2 shows the nonparametric LOD score plots for the change in creatinine clearance. The nonparametric LOD scores (1.68 at 143 cM and 1.05 at 101 cM) were lower than for the cross-sectional analysis of creatinine clearance. The curves seem to follow the pattern and location dictated by the follow-up examination results rather than the baseline results.

Figure 2.

Nonparametric logarithm of odds (NPL LOD) score plots of chromosome 10 markers and changes in creatinine clearance measurements between examinations 1 and 2 () and between examinations 1 and 3 (). 

Full figure and legend (26K)

DISCUSSION

This study analyzed multiple creatinine clearance measurements from a collection of healthy, Caucasian pedigree members. Significant heritability estimates were obtained for creatinine clearance measurements obtained at each of three clinical examinations and also for changes in creatinine clearance between examinations. The genetic underpinnings of creatinine clearance appear to be as strong or stronger for cross-sectional measurements within pedigrees as for longitudinal changes in these measures. While increased measurement error obtained using the difference of two measurements and daily physiological variation could have reduced the heritability estimates of the changes, it appears that there is no clear advantage in studying the creatinine clearance changes to detect underlying associated genes compared to a single examination measurement.

The important result of this study was that significant (P < 0.05) LOD scores were obtained from a linkage analysis of creatinine clearance to markers on chromosome 10 at each of three examinations over a 10-year period. These results replicate previous findings from the rat² and from African American sibpairs with ESRD⁵. The LOD score of 3.4 from the ESRD sample was much higher than that found in this study, probably for multiple reasons. African Americans have a much higher risk of renal failure and ESRD than Caucasians. Using a severe disease end point for analysis, such as ESRD, should increase the accuracy of classification. It may also make the sample more homogeneous. Analyzing a risk factor for development of ESRD, as was done in the current study, may include many subjects who have decreased renal function but who may never develop ESRD.

A genome search of creatinine clearance in the HyperGEN study of hypertension¹⁶ did not find suggestive linkage on chromosome 10 for African American hypertensive siblings (unpublished results). The linkage findings in the rat for Rf-1 indicated that this locus was not linked to blood pressure. Therefore, selection of sibpairs specifically for hypertension may have diluted a linkage signal for creatinine clearance. For Caucasian hypertensive sibpairs in the HyperGEN study, the LOD score for creatinine clearance was 1.4 at 135 cM (unpublished results from¹⁶), despite the selection for hypertension. Even though subjects from the hypertension-ascertained pedigrees in the current study had lower creatinine clearance measurements than those for stroke pedigrees, examination of the individual pedigree LOD scores suggests that most of the linkage signal was derived from the coronary heart disease-ascertained pedigrees that had the lowest mean creatinine clearance. Therefore, these observations suggest that hypertension and stroke may not be closely associated with the underlying gene responsible for the linkage signal. LOD scores in the range of 1.4 to 2.1, as found for analyses of creatinine clearance, are more difficult to replicate in every sample tested than are LOD scores above 3 and more firm conclusions about the association of hypertension or coronary heart disease with this locus and linkage signal differences between Caucasians and African Americans must await further data.

Given the above considerations, it may be remarkable that linkage to the same chromosomal region could be found analyzing creatinine clearance in healthy subjects. If the linkage results from this sample and the ESRD sample represent the same gene predisposing to ESRD, identification of this gene may help predict those who are at increased risk of ESRD.

There are a number of caveats that apply to the findings of this study. The lowest LOD score occurred for examination 1, which theoretically may have been expected to give the highest LOD score because the average of three different 12-hour urine collections was used to calculate creatinine clearance. However, the heritability estimate of creatinine clearance was highest for examination 3, thereby providing a greater chance for a genetic locus to explain a greater proportion of the trait variation, resulting in a higher LOD score. We also speculate that the LOD scores increased over examinations 1 to 3 because of the aging of the pedigree members. Even though children under age 18 at examination 1 were excluded from all analyses, persons at examination 3 were 10 years older than they were at examination 1. Creatinine clearance normally decreases somewhat with age, but if there is a differential effect on that decline by genotype, this difference may be more easily detectable during middle age than during young adulthood. The decrease in creatinine clearance variance explained by gender, age, and BMI from examination 1 to examination 3 indicates less confounding by these variables at older ages. Future multivariate analyses of the longitudinal changes in creatinine clearance may help address these issues.

While each pedigree member had creatinine measured by the same method for each examination, biochemistry autoanalyzers changed among examinations. Therefore, actual estimates of changes using examination 3 data were not possible. Nevertheless, the linkage analysis of the changes should not be affected by changes in methods unless sensitivities of the assay on the two autoanalyzers at the extremes distorted the distributions in some unknown manner. There was no obvious difference among distributions of the three examinations, however, and outlier values were excluded. In addition, the creatinine clearance correlation between examinations 1 and 2 using the same assay was similar to the correlation between examinations 1 and 3 that used different assays. A further caveat is that the standard deviations of the creatinine clearance measurements are somewhat larger than seen in other studies. Since creatinine is fairly stable and straightforward to measure, as is urine volume, the most likely sources of error introduced would be from misreporting of collection times or incomplete sample collection, increasing the variance of the measurements. If this type of error were random, it would probably not introduce spurious linkage signals. One would not expect segregation of errors in nuclear families. The observation that linkage was found using three different clinical examinations and two different assays for urine creatinine reduces the likelihood of spurious linkage. The most likely reason for the larger standard deviations is the selected population of large pedigrees used in this study.

Because of the mild LOD scores detected in this study, identification of the underlying gene would appear to be difficult. The peak LOD scores for examinations 1 and 3 were slightly upstream from the putative location of Rf1, with the examination 2 LOD score closer to the presumed location and coinciding more with the HyperGEN finding¹⁶. Therefore, it is not clear if the underlying gene linked to creatinine clearance is the same as the Rf1 gene. Only a few of the pedigrees in the sample appeared to be responsible for the linkage results and would need to be supplemented with a much larger set of linked pedigrees in order to refine the region. Fortunately, the linkage of renal failure phenotypes in the rat to this region would allow the location of the rat gene to be refined and perhaps identified. The results from the African American sample would also be more amenable for localization of the responsible gene. This gene may then be tested in these pedigrees to verify that it is the gene responsible for the linkage.

References

References

Acknowledgments

This study was supported by grants HL21088, HL24855, HL44738, and AG18734. An abstract discussing these results was presented at the 42nd Annual Conference of Cardiovascular Disease Epidemiology and Prevention, Honolulu, HI, April 23–26, 2002.