Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2016 Apr 28.
Published in final edited form as: Am J Med Genet A. 2005 May 15;135(1):59–65. doi: 10.1002/ajmg.a.30687

Use of the Glycophorin A Somatic Mutation Assay for Rapid, Unambiguous Identification of Fanconi Anemia Homozygotes Regardless of GPA Genotype

Viktoria N Evdokimova 1, Reagan K McLoughlin 2, Sharon L Wenger 3, Stephen G Grant 1,*
PMCID: PMC4849896  NIHMSID: NIHMS765448  PMID: 15822129

Abstract

A 7-year-old girl was hospitalized with pancytopenia requiring blood transfusion. She and an older brother with suspicious symptoms were referred for laboratory testing to confirm a clinical diagnosis of Fanconi anemia (FA). Blood samples from these two children and one parent were examined with the GPA somatic mutation assay. The patient's total GPA somatic mutation frequency of 1.4 × 10−4 was determined despite the confounding effects of her recent transfusion, and was greater than 10-fold higher than that of a population of pediatric controls, consistent with the known FA phenotype. Her brother was not informative for the standard GPA assay, which requires heterozygosity for the MN blood group, but was analyzed with a modified assay that measured only allele loss mutation. His mutation frequency, 6.8 × 10−4 was also supportive of a diagnosis of FA. Both analyses also showed evidence of ongoing mutation through terminal erythroblast differentiation, a characteristic of patients with DNA repair syndromes which further confirmed the diagnoses. These conclusions were confirmed with traditional DEB-induced chromosome breakage studies. The quantitative and qualitative aspects of the GPA assay relevant for applying this test for FA diagnosis, and perhaps for carrier detection, are discussed.

Keywords: Fanconi anemia, FA, Fanconi anemia heterozygotes, GPA, somatic mutations, clinical diagnosis

Introduction

Fanconi anemia (FA) is an autosomal recessive disorder with a monogenic mode of inheritance [Rogatko and Auerbach, 1988], caused by highly heterogeneous mutations in the genes of the FA complementation groups. FA affects all marrow elements, resulting in anemia, leucopenia, and thrombocytopenia, although there is significant clinical diversity [Auerbach et al., 1989]. Leukemia is a fatal complication and may occur in patients lacking full-blown features. FA also has a distinctive cellular phenotype: high frequencies of both spontaneous chromosome breakage and chromosome aberrations, usually rearrangements between non-homologous chromosomes [Schroeder et al., 1964; Bloom et al., 1966; Schroeder and German, 1974]. In addition, lymphocytes from FA patients show a G2 phase arrest [Schindler et al., 1987] and FA cells have accelerated telomere erosion [Callen et al., 2002]. FA cells are hypersensitive to the cytotoxic and clastogenic effects of DNA cross-linking agents, such as the difunctional alkylating agent diepoxybutane (DEB) [Chaganti and Houldsworth, 1991].

The molecular diagnosis of FA is complicated by the heterogeneity of the mutation spectrum and the frequency of intragenic deletions. The accepted diagnostic laboratory test for identification of FA homozygotes remains the in vitro induction of chromosome breakage with DEB or mitomycin C [Auerbach et al., 1979, 1985, 1986; Rosendorff and Bernstein, 1988]. Although this analysis can be performed by most cytogenetics laboratories, it is labor-intensive, involving viable cell culture and some expertise to read the slides.

The flow cytometric glycophorin A (GPA) assay is based on detection and quantitation of somatic “allele loss” mutations at the glycophorin A locus (also called GYPA) on chromosome 4. This locus encodes a major red blood cell surface sialoglycoprotein existing in two common isoforms: M and N [Grant and Bigbee, 1993]. The GPA assay is potentially sensitive to a broad spectrum of mutational events, including point mutation, small insertions and/or deletions, chromosomal aneuploidy, epigenetic gene inactivation, homologous or non-homologous recombination [Grant et al., 1992a]. Previous studies have shown an association between GPA mutation level and elevated risk of cancer [Grant, 2001], particularly in the so-called “DNA repair” diseases ataxia telangiectasia (AT) [Bigbee et al., 1989; Hewitt and Mott, 1992], FA [Bigbee et al., 1991; Sala-Trepat et al., 1993], and Bloom syndrome [Langlois et al., 1987; Kyoizumi et al., 1989b], which show 10-, 50-, and 100-fold increases in GPA mutation frequencies, respectively [Grant et al., 1991]. We have recently shown that this increased GPA mutation frequency, in concert with a characteristic flow cytometric pattern, can be used in the laboratory diagnosis of AT [Grant et al., 1997].

The GPA assay is based on our ability to easily and unambiguously distinguish the two allelic gene products of the GPA locus in GPAM/N heterozygotes and therefore measure mutations with “single-hit” kinetics [Grant et al., 1992a]. This effectively limits the application of the assay to the ∼50% of the population who are heterozygous, reasonable enough for screening studies, but unacceptable for clinical use. This study also demonstrates how the GPA assay may be applied in non-heterozygotes, providing that the expected mutation frequency is higher than background. We also demonstrate how the GPA assay can be applied even when a blood transfusion, a common treatment for anemic conditions, potentially confounds the analysis.

Materials and Methods

Case History

A 7-year-old white female was hospitalized with nosebleed at which time pancytopenia was identified. She received a transfusion of packed red blood cells. Her height and weight were less than the 5th centile. She had a history of fatigue, pallor, poor growth, occasional infections and bruised easily. Her right kidney was somewhat smaller than the left, and malrotated. No other physical findings were noted.

The patient's 8.5-year-old brother had a medical history of hospitalization at 18 months of age for failure to thrive. His height and weight were less than the 5th centile. He had recurrent otitis media, hearing loss, pallor, and was easily fatigued.

A younger brother was apparently normal. The family reported a great maternal uncle who died at age 12, described as “tired.” Blood samples from the patient and her older brother were sent for laboratory confirmation of clinical diagnosis or suspicion of FA, respectively. A sample from one of their parents was also submitted for analysis.

Cytogenetics

Peripheral blood samples from the patient, her older brother, and their parent were processed for routine cytogenetic analysis, including 72 hr incubation in RPMI 1640 medium and trypsin-Giemsa banding [Seabright, 1971]. Blood samples from normal controls were set up concurrently with each patient sample.

For FA testing, DEB was added 24 hr after the initiation of culture to give a final concentration of 0.01 μg/ml in one of two parallel cultures [Auerbach et al., 1981]. Cultures were incubated an additional 48 hr and harvested for metaphases using routine cytogenetic technique. Slides from cultures with and without DEB treatment were solid stained using Giemsa. One hundred cells were scored for chromosomal breakage and interchanges per culture. Laboratory reference for the range of breaks/cell for controls is 0.00–0.10 with or without DEB treatment. For FA patients, the range of spontaneous breakage per cell is 0.05–0.38 and the range of DEB-induced breakage per cell is 0.36–2.12. In our hands, chromosome breakage is 5.4-fold higher in these patients after DEB treatment, which is comparable to the approximately fivefold increase reported in the literature [Auerbach et al., 1981].

Somatic Mutation Assay

The “DB6” version of the GPA somatic mutation assay was performed as described [Grant, 2005]. Intact erythrocytes from the patient, two family members, and controls were “sphered” in hypotonic solution and fixed in formaldehyde/SDS buffer. Sphered cells were then labeled with two monoclonal antibodies, each specific to one allelic isoform of the glycophorin A protein via an extracellular epitope. A standard assay consisted of 5 × 106 cells analyzed by quantitative flow cytometric analysis of the conjugated fluorophors, at flow rates of 3,000–4,000 cells per second, using a rectangular gate in the forward scatter versus log side-scatter distribution to discriminate for intact cells and against antibody induced cell aggregates. Mutation frequencies were calculated as the number of events falling within this defined region of the histogram divided by the total number of analyzed cells. Analysis of samples with unusually high mutation frequencies were repeated, and the results given represent an average of these two assays. Concurrent control blood samples were also obtained from 11 healthy donors: 1 GPAM/N heterozygote and 10 GPAM/M homozygotes. The concurrently analyzed M/N control yielded a total mutation frequency of 8.0 × 10−6, with equal representation of simple allele loss mutants and mutants with a phenotype consistent with allele loss and duplication [Grant et al., 1991]. These results are in excellent agreement with the historic mutation frequencies associated with this individual, who has been assayed 25 times in our laboratory, with mean total GPA mutation frequencies of 9.4 ± 1.0 × 10−6 cells analyzed, allele loss frequencies of 4.3 ± 0.4 × 10−6, and loss and duplication frequencies of 5.1 ± 0.8 × 10−6 (P ≈ 0.4 for all three measurements). Historical control samples for this study were obtained from employees of the Lawrence Livermore National Laboratory and their families, from 1988 through 1991. Subsets of this data have been previously reported [Jensen et al., 1987, 1991]. Reference samples from confirmed FA patients and obligate heterozygotes were obtained from Dr. Arlene D. Auerbach of the International Fanconi Anemia Registry (IFAR) [Kutler et al., 2003].

Statistical Analysis

Analysis of individual results in the context of control populations was done with the z-test at ∝=0.05 on ln transformed data. Comparisons between populations were performed with the t-test at the same level of significance on similarly transformed data.

Results

Cytogenetics

The hospitalized patient had a normal female karyotype. She had 7 spontaneous gaps and breaks and one quadriradial out of 50 cells, and 75 breakage/interchange configurations among 100 cells following treatment with DEB. The latter included chromosomal gaps and breaks, a dicentric, triradials, quadriradials, acentric fragments and multiple interchange figures. Both of these values are well above the range of normal controls in our hands, and confirm a diagnosis of FA.

The older brother had a normal male karyotype and spontaneous gaps and breaks in 14 out of 100 cells. His DEB-induced culture contained 53 breakage/interchange configurations among 100 cells, which included gaps, breaks, a triradial, quadriradials, and acentric fragments. A bone marrow aspirate drawn at the time of testing was hypocellular. Once again, these results confirm the clinical diagnosis of FA.

The parent's spontaneous chromosome breakage level was 5/100 cells, which increased to 11/100 in the DEB-treated culture. The latter is slightly higher than the normal range of 1–10 induced breakage/interchange configurations per 100 cells, which may be related to his carrier status of the FA gene, but cannot be used reliably to diagnose heterozygotes [Cohen et al., 1982; Cervenka and Hirsch, 1983].

Somatic Mutation Analysis

Following routine cytogenetic and DEB tests, 500 μl aliquots of the whole blood samples from the patient, her brother, their parent, and healthy controls were serotyped for the M/N blood group and further processed for analysis with the GPA somatic mutation assay. The GPA assay enumerates variant cells potentially arising by a wide variety of molecular mechanisms associated with allelic “segregation” or loss of heterozygosity (LOH) [Worton and Grant, 1985; Grant et al., 1991]. These mutants can be further characterized by the level of expression of the remaining GPA allele into “N/Ø” variants, with a simple allele loss phenotype, and “N/N” cells, which, in addition to the loss of the M allele, express the remaining N allele at twice the normal level. Previous studies have demonstrated an increased frequency of uninduced GPA mutation in FA patients corresponding to an average 30-fold increase in N/Ø variants and an 8-fold increase in N/N variants compared to unaffected controls [Bigbee et al., 1991; Sala-Trepat et al., 1993].

The patient had a heterozygous M/N phenotype, ideal for evaluation by the GPA assay. In practice, however, the flow histogram of her sample revealed a more complicated situation (Fig. 1). Major peaks of approximately equal size were evident in the areas of the histogram associated with both an M/M and an M/N genotype. It was hypothesized that one of the peaks consisted of cells from a transfusion of packed cells given 9 days prior to sample acquisition. Quantitation of the nucleic acid content of circulating red cells by staining with propidium iodide (PI) revealed that the cells comprising the M/M peak had a very low proportion of cells retaining nucleic acid, specifically mRNAs in reticulocytes, of 0.3% [Corver et al., 1994]. Cells in the M/N peak, and variant cells falling in the allele loss window had much higher reticulocyte proportions of 4.7%, suggesting that they were the patient's own cells, and that the vast majority of M/M cells were derived from the transfusion. The total GPA mutation frequency of the patient was 71.9 × 10−6 (P = 0.0024 versus normal pediatric controls), with an allele loss (N/Ø) frequency of 59.7 × 10−6 (P < 0.0001), and an allele loss and duplication (N/N) frequency of 12.2 × 10−6 (P = 0.20). This N/Ø frequency is already high enough to be considered an “outlier” [Bigbee et al., 1998] and is supportive of a diagnosis of FA [Bigbee et al., 1991; Sala-Trepat et al., 1993; Grant et al., 1997], but the analysis becomes even more definitive if these frequencies are adjusted for the number of transfused cells confounding the assay: total GPA mutation frequency = 143.5 × 10−6 (P < 0.0001), N/Ø frequency = 119.2 × 10−6 (P < 0.0001), N/N frequency = 24.3 × 10−6 (P = 0.041). The allele loss frequency from this patient is compared to those from the two previously reported, largely pediatric populations and a matched set of controls in Figure 2.

Fig. 1.

Fig. 1

Open in a new tab

Flow cytometric histograms of 106 erythrocytes from a normal control (A), a confirmed Fanconi anemia (FA) patient (B), and the current patient (C) analyzed with the GPA assay. The major peak in all three consists of heterozygous M/N cells with equal labeling with both fluorophors. Allele loss mutations appear in a window directly beneath the main peak with less than 1% GPA(M) fluorescence. Loss and duplication mutations appear in a equally sized window directly to the right of the allele loss window, indicative of a twofold increase in GPA(N) fluorescence (note log scale). Together, these two windows define total GPA mutation. A “ridge” of cells joining the main peak to the mutant window is evident in both the affected FA patient and the diagnostic sample. An “arm” of N-allele loss and loss and duplication mutants is also evident extending left from the main peak in the FA patient, but this area of the distribution is obscured in our patient sample by a large peak of M/M transfused cells.

Fig. 2.

Fig. 2

Open in a new tab

Comparison of GPA allele loss mutation frequencies among FA patients, controls and diagnostic samples. Known FA patients and age-matched controls were derived from unpublished data and that of Sala-Trepat et al. [1993]. All are GPAM/N heterozygotes, as is the patient. M/M controls were obtained specifically for comparison with the patient's brother, but as adult samples, would be expected to have a slightly higher mutation frequency. These samples also reflect allele-loss from two alleles rather than one, and may have a higher background simply because they cannot be distinguished from the constitutive GPA phenotype as well by the antibodies used in the GPA assay.

Although the uninduced N/Ø mutation frequency for this patient is elevated and therefore is consistent with a diagnosis of FA, this observation is not, in itself, conclusive (although in conjunction with the clinical findings reported above, it may be considered confirmative). Individuals with similarly elevated mutation frequencies have been observed in all populations surveyed with the GPA assay [Jensen et al., 1987; Akiyama et al., 1995; Radack et al., 1996]. We have found that the proportion of such outlier individuals is ∼2% in newborns [Manchester et al., 1995], and increases with age [Bigbee et al., 1998]. In our previous application of the GPA assay to the diagnosis of AT [Grant et al., 1997], we pointed out several characteristics of the flow histogram that help in discriminating these patients from outlying controls, and some of these are also applicable to FA. First, we noted that the elevation in GPA mutation in AT was due exclusively to generation of allele loss type mutants; there was no evidence for an effect on the frequency of loss and duplication mutation in that disease. As mentioned earlier, it has been reported that, while an elevation in allele loss mutation is also a consistent observation in FA, there is also evidence for an elevation in loss and duplication mutants, although of lesser magnitude [Sala-Trepat et al., 1993]. In our hands, any elevation in the frequency of loss and duplication mutants is highly variable, and we do not consider it to be a reliable feature of the analysis [Grant et al., 2004]. In particular, it must be assured that additional events in the N/N window represent a true peak of variant cells with a loss and duplication phenotype, and are not simply spill over events from the large peak of N/Ø variants. In any case, an increased frequency of loss and duplication mutants may be considered as a possible but not mandatory feature of the FA phenotype. Our patient had a slightly elevated N/N mutation frequency when compared to those of the control population given in Figure 2 (7.7 ± 1.0 × 10−6), but this does not weigh significantly in the diagnostic efficacy of her GPA analysis.

The next feature that typifies the GPA analysis of an AT or FA patient is the presence of a peak of cells corresponding to loss of the N allele rather than the M allele. Our preference for quantification of M allele loss in the form of the N/Ø and N/N variant classes is due to the technical parameters of the detection system rather than any inherent genetic difference in the M and N alleles. The N/Ø and N/N windows are set at less than 1% M allele-specific fluorescence, usually assuring unambiguous discrimination of wild type and mutant cells. The N-specific antibody used in the assay, however, cross-reacts with another, related sialoglycoprotein on the red cells surface, glycophorin B, which is expressed at about 1/3 the level of glycophorin A [Cartron and Rahuel, 1995]. Thus, loss of signal from a single copy of the GPA N allele reduces the total fluorescein fluorescence by less than an order of magnitude, rather than the 100-fold loss of phycoerythrin fluorescence associated with loss of an M allele. We, therefore, only quantify N allele-loss variants in cases such as this, where a large peak of mutants is expected (others regularly quantitate N allele-loss, however, in their unique version of the GPA assay [Kyoizumi et al., 1989a]). In any case, if FA is characterized by high frequencies of ongoing mutation, there should be approximately equal peaks of M allele loss and N allele loss mutants [Grant et al., 1997]. In this patient, however, the large peak of M/M cells attributable to transfusion obscures this area of the histogram. The fact that there were any PI-positive reticulocytes detectable in this peak, when the transfusion occurred over a week prior to sampling might suggest that there are actually a considerable number of mutant M/Ø and M/M variants generated by the patient contributing to this peak.

Finally, the most powerful qualitative feature of the flow histogram from an AT or FA patient is the presence of a “ridge” of cells joining the allele loss mutant peaks with the wild-type peaks [Grant et al., 1997]. This ridge does not appear in GPA analyses of normal individuals, even of outliers with unusually high mutation frequencies [Bigbee et al., 1998], nor does it appear in analyses of individuals with high mutation frequencies due to historic genotoxic exposures, such as survivors of the atomic bombing of Hiroshima [Langlois et al., 1987; Kyoizumi et al., 1989a]. We believe that this ridge is due to high frequencies of ongoing mutation occurring after the onset of GPA expression during erythroblast maturation, leading to a spectrum of partial mutant phenotypes. Besides patients with DNA repair syndromes, the only other situation where we have seen this phenomenon is in patients with an ongoing significant genotoxic exposure, such as cancer patients undergoing mutagenic chemotherapy [Bigbee et al., 1990; Grant et al., 2004]. This patient clearly shows evidence of such ongoing mutation (Fig. 2). Once again, we would expect a similar ridge linking the wild type peak to the N-allele loss peak, if it were not obscured by the transfused cells.

The patient's older brother displayed some suspicious clinical symptoms, so it was even more important to provide laboratory data on whether he might have FA. Unfortunately, his GPA phenotype was homozygous M/M. In such a case, it is completely impossible to observe allele loss plus duplication events, since it regenerates the original genotype. Simple allele loss, to produce an M/Ø phenotype, should occur, but it would only reduce the M-specific labeling by half, and these variants would occur in the “apron” of the main M/M peak. On the other hand, since there are two, indistinguishable M alleles in this individual, the frequency of M-allele loss variants should be twice as high as that observed in M/N heterozygotes. In running this sample, we hypothesized that if there was a large enough peak of M/Ø cells, suggesting that the brother was affected, it would be distinguishable as a separate peak immediately below the main M/M peak (Fig. 3).

Fig. 3.

Fig. 3

Open in a new tab

Flow cytometric histograms of 106 erythrocytes from a normal GPAM/M control (A), and the current patient's brother (B) analyzed with the GPA assay. The major peak in both panels consists of homozygous M/M cells. This peak is only 1/3 closer to the Y-axis than the M/N peaks seen in Figure 1 because of cross-reaction of our GPA(N)-specific antibody with glycophorin B on the red cell surface. Allele loss phenotype M/Ø cells would fall directly beneath the main peak, with only half the GPA(M)-specific fluorescence of the main peak. At low, normal mutation frequencies, these cells (∼10–20) are lost in the apron of the main peak. In the diagnostic sample, however, an unambiguous M/Ø is observed, as well as a ridge of cells with intermediate phenotypes extending to the main peak.

Indeed, such a distinct peak was observed in the GPA flow histogram of this individual, yielding a frequency of allele-loss variants of 1360.6 × 10−6. This is clearly an abnormally high “outlier” frequency when compared to our N/Ø variant frequencies from M/N heterozygotes, even when it is normalized to a single allele by dividing by the number, two, of M alleles that can be “lost” (= 680.3 × 10−6 [P < 0.0001]). There is also evidence for a ridge joining the mutant and wild-type peaks, an important qualitative feature of AT and FA GPA analyses. Since this analysis seemed to be useful, we similarly analyzed 10 samples from normal M/M individuals (Fig. 2; M/Ø variant frequency = 31.9 ± 4.8 × 10−6), clearly establishing that the patient's brother had an unusually high frequency of allele loss mutation (>20-fold higher than controls, despite the fact that these are adult, not pediatric controls, and there is an established age effect in GPA somatic mutation [Jensen et al., 1987; Radack et al., 1996], even when performed with the assay that includes M/Ø variants in their analysis [Akiyama et al., 1995]). We therefore concluded that the results of the GPA analysis on this individual were supportive of a diagnosis of FA.

Based on these diagnoses, the children's parent must be an obligate heterozygote for FA. They had a rather low total GPA mutation frequency of 12.0 × 10−6 (P = 0.43 versus normal adult controls), that was fairly equally divided between the allele loss (6.4 × 10−6 [P = 0.34]) and loss and duplication (5.6 × 10−6 [P = 0.44]) classes. Indeed, this total GPA mutation frequency is significantly lower (P = 0.033) than that of eight such obligate heterozygotes reported in Sala-Trepat et al. [1993] (Table I), who concluded that FA heterozygotes have slightly but significantly elevated GPA mutation frequencies. The low frequency in the patients' parent is primarily due to the loss and duplication class of variant (P = 0.005), since there is no significant difference in the frequency of allele loss variants (P = 0.39). If the comparison is extended to include a set of 28 obligate heterozygotes identified by the International Fanconi Anemia Registry (IFAR) at Rockefeller University, however, the present sample does not differ from this population in total GPA mutation frequency (P = 0.18), allele loss frequency (P = 0.35), or loss and duplication frequency (P = 0.17) (data not shown).

Table I. GPA Mutation Frequencies in Fanconi Anemia (FA) Heterozygote and Control Populations.

N Total GPA mutation frequenciesa
N/Ø N/N
FA heterozygote 1 12.0 6.4 5.6
FA heterozygotesb 8 26.5 ± 4.0 13.5 ± 4.9 13.0 ± 1.2
Controls (aged 4–24)b 21 13.0 ± 1.4 6.2 ± 0.8 6.9 ± 0.7
Controls (aged 8–24) 61 13.5 ± 1.4 5.3 ± 0.4 8.2 ± 1.3
Controls (aged 25–55) 257 17.0 ± 0.6 7.2 ± 0.4 9.8 ± 0.5
Open in a new tab
a

× 10−6.

Discussion

The data presented in this article confirm the reported elevated levels of somatic mutation at the autosomal GPA locus in FA patients and demonstrate how this result can be applied diagnostically. To date, we have performed a total of 152 analyses on putative FA patient samples and, by applying the diagnostic criteria outlined in the results, we have a perfect concordance with the DEB-induced chromosome breakage assay. This includes 145 samples from the IFAR (including 36 known FA patients) and 7 diagnostic cases, 3 positive, 4 negative. Thus, the GPA assay may be used either in conjunction with the DEB assay, or even instead of the DEB assay. The GPA assay requires much less blood than the DEB assay (essentially 100 μl, since phenotyping is no longer required), which is important in evaluating pediatric cases, requires no cell culturing, in vitro exposure or cytogenetic evaluation, and can be performed in a matter of hours. We have already shown that the GPA assay can be used to diagnose AT [Grant et al., 1997] and application to Bloom syndrome should be straightforward [Kyoizumi et al., 1989a; Langlois et al., 1989]. Thus, the assay has multiple clinical uses. Elevated GPA somatic mutation frequencies have also been observed in patients with Nijmegan breakage syndrome [Grant et al., 1992b], Cockayne syndrome [Lin et al., 1995], Werner syndrome [Kyoizumi et al., 1998; Moser et al., 2000], and possibly Rothmund–Thompson syndrome [Grant et al., 2000]. GPA mutation frequencies are also elevated in cancer patients [Okada et al., 1997; Grant, 2001]. Most significantly, we have shown that the GPA assay can be used to quantitatively biomonitor the effects of genotoxic chemotherapy [Bigbee et al., 1989; Grant et al., 2004], once again, regardless of GPA genotype, so that it may become a standard clinical technique.

We have previously published a set of guidelines for applying the GPA assay for the diagnosis of AT [Grant et al., 1997]. A similar set of criteria can be developed for FA. First, since allele loss mutation frequencies in FA patients are typically fivefold higher than the already elevated levels observed in AT [Grant et al., 1992a,b], similar quantitative criteria can be adopted for the two diseases, although the test should display a greater sensitivity for FA. AT patients have shown no evidence of increased frequencies of loss and duplication events, whereas there seems to be a variable increase in this class of mutants in FA. We suggest that this element of the analysis should be considered as supportive of the diagnosis if it is elevated, but neutral if it is not present. The best way to combine these criteria is to consider the total GPA mutation frequency as well as the N/Ø mutation frequency. Thus, a putative FA patient should have either an allele loss frequency equal to or greater than 30 × 10–6, or a total GPA mutation frequency equal to or greater than 50 × 10–6, or both. Applying these criteria to the combined FA datasets and younger (≤24-years-old) controls in Table I, the allele loss aspect of the test alone has a sensitivity of 92% and a specificity of 97% (four false negatives and one false positive). The contribution of the loss and duplication mutants is demonstrated by the improvement when the total GPA mutation frequency criterion is applied to this population: the test now has a sensitivity of 98% and a specificity of 100%. Indeed, the one remaining false negative has a combined mutation frequency of 49.4 × 10−6, whereas the highest control has a frequency of 28.0 × 10−6. If older adults and FA heterozygotes are added to the test population the sensitivity is unaffected, and the specificity is retained at a high level (98%), but there are now 10 false positives by either testing criteria. This illustrates the need to add qualitative assessment of the histogram to the testing procedure, especially if it is applied to adults. It is interesting to note that 3 and 4, respectively, of the false positives for allele loss and total GPA mutation frequency were FA heterozygotes.

The GPA assay was originally developed as a screening assay for human genotoxic exposures, and in this context it has been widely applied [Grant and Bigbee, 1993]. The genetic limitation that only GPAM/N heterozygotes are informative for the test has not been a major detriment to its application in exposure monitoring, since that is often a population-based procedure. For clinical application, however, having the test applicable to only half the test population has been an insurmountable problem. For example, in the populations presented in this article, a total of 192 samples received from IFAR (including known FA homozygotes, family members, and controls) were not assayed because of their homozygous GPA genotype. In addition, six diagnostic samples were not analyzed for the same reason. The present data suggest that all of these samples are now approachable from the standpoint of determining whether or not they are indicative of the FA phenotype. In this study, we report the diagnosis of an FA homozygote with an M/M GPA phenotype, because allele loss mutants, the major class affected by FA, are still detectable in such individuals (indeed, since they have two M alleles, their frequency is actually enhanced twofold over M/N heterozygotes). Data from cancer patients, where the mutants are induced by genotoxic chemotherapy, suggest that similar results are possible from N/N homozygotes.

An additional 24 known FA homozygous samples from IFAR were not scored for GPA mutation due to recent transfusions. As this is a common treatment for the pancytopenia associated with FA (and, indeed, was the reason the current patient was referred for testing), any widespread diagnostic application of the GPA assay in putative FA patients depends on our ability to overcome this problem. We were fortunate in this case that the transfusion was comprised only of cells of M/M genotype, in a patient with a distinct M/N genotype. Thus, although the transfused red cells obscured any N-allele loss or loss and duplication variant cells, the presence of these cells did not affect the interpretation of the assay. Had the transfusion been of cells of mixed GPA phenotypes, e.g., from multiple donors, it is possible that the analysis would not have been informative. If the transfusion had been of N/N cells, these cells would have obscured the major peak of mutants used to establish the FA diagnosis, although, in this case, the N-allele loss peak (with an M/Ø phenotype) would not have been obscured. In some cases, with sufficient clinical support, this might be enough to confirm a diagnosis of FA. If the transfused cells had been of the same genotype as the patient, heterozygous M/N, these cells would have contributed greatly to the “normal” peak and only at background “normal” frequencies to the variants, effectively diluting the mutational signal used for diagnosis. Moreover, the effect of the transfused cells on the denominator of the calculation, total cells analyzed, cannot be compensated for, as was done for our patient, since the two cell populations would be indistinguishable. If the transfusion contributed approximately half of the red cell population, as was seen in the current patient, this dilution would reduce the sensitivity of the GPA-based diagnostic test to 84%, based on our current population. Thus, if the GPA analysis is to be performed on an M/N individual, it is best to either obtain the sample before any transfusions or to take this intent into account in the choice of transfused cells. Thus, for M/N heterozygous patients, M/M transfused cells are the least intrusive in the analysis. Similarly, for M/M and N/N patients, transfusions of cells with the opposite homozygous phenotype would be best.

The parents in this family are obligate heterozygotes for FA, but the GPA assay results from the parent who provided a blood sample were both quantitatively and qualitatively normal (although their DEB-induced chromosome breakage frequency was slightly high). The chromosome breakage assay has been proposed for use for carrier detection [Auerbach and Wolman, 1978; Auerbach et al., 1981], however, it is not considered either reliable or unequivocal in this context [Cohen et al., 1982; Cervenka and Hirsch, 1983]. Sala-Trepat et al. [1993] reported GPA mutation frequencies on eight obligate heterozygotes (Table I). The total GPA mutation frequency, as well as both the N/Ø and N/N frequencies of the FA heterozygotes were significantly higher than their concurrently analyzed normal controls (P < 0.001, 0.03, and <0.001, respectively), however, the value of this comparison was diminished by the difference in ages between the two populations (the oldest control was younger than the youngest heterozygote). If their data from FA heterozygotes is compared to a subset of our own healthy control database with the same age range, both total GPA mutation frequency (P = 0.008) and N/Ø frequency (P = 0.01) are significantly elevated (Table I). Although it is unusual to compare populations across laboratories like this, the validity of the result is supported by the lack of difference observed when comparing control populations of similar age range (P = 0.85, 0.25, and 0.55, for total mutation frequency, allele loss and loss and duplication, respectively). In addition, we found that FA heterozygotes were much more likely to appear as false positives in our diagnostic GPA analysis. Indeed, the odds ratios (ORs) for FA heterozygotes with mutation frequencies above the diagnostic criteria were 8.04 (95% confidence interval [CI] 3.42–18.9) for total GPA mutation, 5.00 (95% CI 0.91–27.2) for allele loss alone, and 6.17 (95% CI 4.00–9.53) when the two criteria are combined (comparable ORs for FA homozygotes under the same circumstances range from 476 to 3,450). These criteria detect only 14% of FA heterozygotes, however, and 2.6% of normals. The OR for FA heterozygote detection by total GPA mutation frequency peaks at 9.77 (95% CI 7.10–13.4) at a cut-off of 40.0 × 10−6, which captures 26% of heterozygotes, and 3.4% of controls, but remains positive and significant down to a cut-off of 10.0 × 10−6, which would capture all heterozygotes including our patient's parent, but has an unacceptable false positive rate of 70% of the normal population.

Thus, these data suggest that FA heterozygotes may indeed have a slightly increased mutation frequency, mainly characterized by simple allele loss, compared to controls, but that the magnitude of the difference is unlikely to allow for unequivocal identification of individual heterozygotes. These data do, however, support the concept that FA heterozygotes have a distinct phenotype that predisposes them to cancer [Heim et al., 1992; Djuzenova et al., 2001]. This was unambiguously established when it was discovered that BRCA2, a cancer-predisposing tumor suppressor gene in heterozygous mutation carriers, was identified as the FANCD1 gene, causing FA in the homozygous inactive state [Howlett et al., 2002].

Acknowledgments

We would like to thank Dr. Arlene Auerbach and the staff of IFAR for providing samples from known FA patients, their families, and controls. The technical expertise of Barbara A. Nisbet, Ann E. Gorvad, Britt L. Luccy, and Christina M. Cerceo is gratefully acknowledged.

Grant sponsor: NIH; Grant number: HD33016; Grant sponsor: University of Pittsburgh (Competitive Medical Research grant).

References

  1. Akiyama M, Kyoizumi S, Hirai Y, Kusonoki Y, Iwamoto KS, Nakamura N. Mutation frequency in human blood cells increases with age. Mutat Res. 1995;338:141–149. doi: 10.1016/0921-8734(95)00019-3. [DOI] [PubMed] [Google Scholar]
  2. Auerbach AD, Wolman SR. Carcinogen-induced chromosome breakage in Fanconi's anemia heterozygous cells. Nature. 1978;271:69–71. doi: 10.1038/271069a0. [DOI] [PubMed] [Google Scholar]
  3. Auerbach AD, Warburton D, Bloom AD, Chaganti RSK. Prenatal detection of the Fanconi's anemia gene by cytogenetic methods. Am J Hum Genet. 1979;31:77–81. [PMC free article] [PubMed] [Google Scholar]
  4. Auerbach AD, Adler B, Chaganti RS. Prenatal and postnatal diagnosis and carrier detection of Fanconi anemia by a cytogenetic method. Pediatrics. 1981;67:128–135. [PubMed] [Google Scholar]
  5. Auerbach AD, Sagi M, Adler B. Fanconi anemia: Prenatal diagnosis in 30 fetuses at risk. Pediatrics. 1985;76:794–800. [PubMed] [Google Scholar]
  6. Auerbach AD, Min Z, Ghosh R, Pergament E, Verlinsky Y, Nicolas H, Boue J. Clastogen-induced chromosomal breakage as a marker for first trimester prenatal diagnosis of Fanconi anemia. Hum Genet. 1986;73:86–88. doi: 10.1007/BF00292671. [DOI] [PubMed] [Google Scholar]
  7. Auerbach AD, Rogatko A, Schroeder-Kurth TM. International Fanconi Anemia Registry: Relation of clinical symptoms to diepoxybutane sensitivity. Blood. 1989;73:391–396. [PubMed] [Google Scholar]
  8. Bigbee WL, Langlois RG, Swift M, Jensen RH. Evidence for an elevated frequency of in vivo somatic cell mutations in ataxia telangiectasia. Am J Hum Genet. 1989;44:402–408. [PMC free article] [PubMed] [Google Scholar]
  9. Bigbee WL, Wyrobek AW, Langlois RG, Jensen RH, Everson RB. The effect of chemotherapy on the in vivo frequency of glycophorin A ‘null’ variant erythrocytes. Mutat Res. 1990;240:165–175. doi: 10.1016/0165-1218(90)90056-8. [DOI] [PubMed] [Google Scholar]
  10. Bigbee WL, Jensen RH, Grant SG, Langlois RG, Olsen DA, Auerbach AD. Evidence for elevated in vivo somatic mutation at the glycophorin A locus in Fanconi anemia. Am J Hum Genet. 1991;49(Suppl):446. [Google Scholar]
  11. Bigbee WL, Fuscoe JC, Grant SG, Jones IM, Gorvad AE, Harrington-Brock K, Strout CL, Thomas CB, Moore MM. Human in vivo somatic mutation measured at two loci: Individuals with stably elevated background erythrocyte glycophorin A (gpa) variant frequencies exhibit normal T-lymphocyte hprt mutant frequencies. Mutat Res. 1998;397:119–136. doi: 10.1016/s0027-5107(97)00186-3. [DOI] [PubMed] [Google Scholar]
  12. Bloom GE, Warner S, Gerald PS, Diamond LK. Chromosome abnormalities in constitutional aplastic anemia. N Eng J Med. 1966;274:8–14. doi: 10.1056/NEJM196601062740102. [DOI] [PubMed] [Google Scholar]
  13. Callen E, Samper E, Ramirez MJ, Creus A, Marcos R, Ortega JJ, Olive T, Badell I, Blasco MA, Surralles J. Breaks at telomeres and TRF2-independent end fusions in Fanconi anemia. Hum Mol Genet. 2002;11:439–444. doi: 10.1093/hmg/11.4.439. [DOI] [PubMed] [Google Scholar]
  14. Cartron JP, Rahuel C. MNSs and major glycophorins of human erythrocytes. Transfus Clin Biol. 1995;2:251–258. doi: 10.1016/s1246-7820(05)80091-6. [DOI] [PubMed] [Google Scholar]
  15. Cervenka J, Hirsch BA. Cytogenetic differentiation of Fanconi anemia, “idiopathic” aplastic anemia, and Fanconi anemia heterozygotes. Am J Med Genet. 1983;15:211–223. doi: 10.1002/ajmg.1320150205. [DOI] [PubMed] [Google Scholar]
  16. Chaganti RSK, Houldsworth J. Fanconi anemia: A pleiotropic mutation with multiple cellular and developmental abnormalities. Ann Genet. 1991;34:206–211. [PubMed] [Google Scholar]
  17. Cohen MM, Simpson SJ, Honig GR, Maurer HS, Nicklas JW, Martin AO. The identification of Fanconi anemia genotypes by clastogenic stress. Am J Hum Genet. 1982;34:794–810. [PMC free article] [PubMed] [Google Scholar]
  18. Corver WE, Cornelisse CJ, Fleuren GJ. Simultaneous measurementof two cellular antigens and DNA using fluorescein-isothiocyanate, R-phycoerythrin, and propidium iodide on a standard FACScan. Cytometry. 1994;15:117–128. doi: 10.1002/cyto.990150205. [DOI] [PubMed] [Google Scholar]
  19. Djuzenova CS, Rothfuss A, Oppitz U, Speit G, Schindler D, Hoehn H, Flentje M. Response to X-irradiation of Fanconi anemia homozygous and heterozygous cells assessed by the single-cell gel electrophoresis (comet) assay. Lab Invest. 2001;81:185–192. doi: 10.1038/labinvest.3780226. [DOI] [PubMed] [Google Scholar]
  20. Grant SG. Molecular epidemiology of human cancer: Biomarkers of genotoxic exposure and susceptibility. J Environ Pathol Toxicol Oncol. 2001;20:245–261. [PubMed] [Google Scholar]
  21. Grant SG. The GPA in vivo somatic mutation assay. Meth Mol Biol. 2005;291:179–195. doi: 10.1385/1-59259-840-4:179. [DOI] [PubMed] [Google Scholar]
  22. Grant SG, Bigbee WL. In vivo somatic mutation and segregation at the human glycophorin A (GPA) locus: Phenotypic variation encompassing both gene-specific and chromosomal mechanisms. Mutat Res. 1993;288:163–172. doi: 10.1016/0027-5107(93)90217-4. [DOI] [PubMed] [Google Scholar]
  23. Grant SG, Bigbee WL, Langlois RG, Jensen RH. Allele loss at the human GPA locus: A model for recessive oncogenesis with potential clinical application. Clin Biotechnol. 1991;3:177–185. [Google Scholar]
  24. Grant SG, Bigbee WL, Langlois RG, Jensen RH. Methods for the detection of somatic mutational and segregational events: Relevance to the monitoring of survivors of childhood cancer. In: Green DM, D'Angio GJ, editors. Late effects of treatment for childhood cancer. New York: Wiley-Liss; 1992a. pp. 133–150. [Google Scholar]
  25. Grant SG, Langlois RG, Auerbach AD, Manchester DK, Monnat RJ, Jensen RH, Bigbee WL. Analysis of human DNA metabolism/repair diseases with the in vivo glycophorin A somatic segregation assay: Application of an improved assay and extension to new syndromes. Environ Mol Mutagen. 1992b;19(Suppl 20):20. [Google Scholar]
  26. Grant SG, Reeger W, Wenger SL. Diagnosis of ataxia telangiectasia with the glycophorin A somatic mutation assay. Genet Testing. 1997;1:261–267. doi: 10.1089/gte.1997.1.261. [DOI] [PubMed] [Google Scholar]
  27. Grant SG, Wenger SL, Latimer JJ, Thull D, Burke LW. Analysis of genomic instability using multiple assays in a patient with Rothmund–Thomson syndrome. Clin Genet. 2000;58:209–215. doi: 10.1034/j.1399-0004.2000.580308.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Grant SG, Wenger SL, Rubinstein WS, Latimer JJ, Bigbee WL, Auerbach AD. Elevated levels of somatic mutation in homozygotes and heterozygotes for inactivating mutations in the genes of the FA/BRCA DNA repair pathway. Am J Hum Genet. 2004;75(Suppl):94. [Google Scholar]
  29. Heim RA, Lench NJ, Swift M. Heterozygous manifestations in four autosomal recessive human cancer-prone syndromes: Ataxia telangiectasia, xeroderma pigmentosum, Fanconi anemia, and Bloom syndrome. Mutat Res. 1992;284:25–36. doi: 10.1016/0027-5107(92)90022-t. [DOI] [PubMed] [Google Scholar]
  30. Hewitt M, Mott MG. The assessment of in vivo somatic mutations in survivors of childhood malignancy. Br J Cancer. 1992;66:143–147. doi: 10.1038/bjc.1992.232. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Howlett NG, Taniguchi T, Olson S, Cox B, Waisfisz Q, De Die-Smulders C, Persky N, Grompe M, Joenje H, Pals G, Ikeda H, Fox EA, D'Andrea AD. Biallelic inactivation of BRCA2 in Fanconi anemia. Science. 2002;297:606–609. doi: 10.1126/science.1073834. [DOI] [PubMed] [Google Scholar]
  32. Jensen RH, Bigbee WL, Langlois RG. In vivo somatic mutations in the glycophorin A locus of human erythroid cells. Banbury Rep. 1987;28:149–159. [Google Scholar]
  33. Jensen RH, Grant SG, Langlois RG, Bigbee WL. Somatic cell genotoxicity at the glycophorin A locus in humans. Prog Clin Biol Res. 1991;372:329–339. [PubMed] [Google Scholar]
  34. Kutler DI, Singh B, Satagopan J, Batish SD, Berwick M, Giampietro PF, Hanenberg H, Auerbach AD. A 20-year perspective on the International Fanconi Anemia Registry (IFAR) Blood. 2003;101:1249–1256. doi: 10.1182/blood-2002-07-2170. [DOI] [PubMed] [Google Scholar]
  35. Kyoizumi S, Nakamura N, Hakoda M, Awa AA, Bean MA, Jensen RH, Akiyama M. Detection of somatic mutations at the glycophorin A locus in erythrocytes of atomic bomb survivors using a single beam flow sorter. Cancer Res. 1989a;49:581–588. [PubMed] [Google Scholar]
  36. Kyoizumi S, Nakamura N, Takebe H, Tatsumi K, German J, Akiyama M. Frequency of variant erythrocytes at the glycophorin-A locus in two Bloom's syndrome patients. Mutat Res. 1989b;214:215–222. doi: 10.1016/0027-5107(89)90166-8. [DOI] [PubMed] [Google Scholar]
  37. Kyoizumi S, Kusunoki Y, Seyama T, Hatamochi A, Goto M. In vivo somatic mutations in Werner's syndrome. Hum Genet. 1998;103:405–410. doi: 10.1007/s004390050841. [DOI] [PubMed] [Google Scholar]
  38. Langlois RG, Bigbee WL, Kyoizumi S, Nakamura N, Bean MA, Akiyama M, Jensen RH. Evidence for increased somatic cell mutations at the glycophorin A locus in atomic bomb survivors. Science. 1987;236:445–448. doi: 10.1126/science.3563520. [DOI] [PubMed] [Google Scholar]
  39. Langlois RG, Bigbee WL, Jensen RH, German J. Evidence for elevated in vivo mutations and somatic recombination in Bloom's syndrome. Proc Natl Acad Sci USA. 1989;86:670–674. doi: 10.1073/pnas.86.2.670. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Lin YW, Kubota M, Hirota H, Furusho K, Tomiwa K, Ochi J, Kasahara Y, Sasaki H, Ohta S. Somatic cell mutation frequency at the HPRT, T-cell antigen receptor and glycophorin A loci in Cockayne syndrome. Mutat Res. 1995;337:49–55. doi: 10.1016/0921-8777(95)00014-b. [DOI] [PubMed] [Google Scholar]
  41. Manchester DK, Nicklas JA, O'Neill JP, Lippert MJ, Grant SG, Langlois RG, Moore DH, III, Jensen RH, Albertini RJ, Bigbee WL. Sensitivity of somatic mutations in human umbilical cord blood to maternal environments. Environ Mol Mutagen. 1995;26:203–212. doi: 10.1002/em.2850260304. [DOI] [PubMed] [Google Scholar]
  42. Moser MJ, Bigbee WL, Grant SG, Emond MJ, Langlois RG, Jensen RH, Oshima J, Monnat RJ., Jr Genetic instability and hematologic disease risk in Werner syndrome patients and heterozygotes. Cancer Res. 2000;60:2492–2496. [PubMed] [Google Scholar]
  43. Okada S, Ishii H, Nose H, Okusaka T, Kyogoku A, Yoshimori M, Wakabayashi K. Evidence for increased somatic cell mutations in patients with hepatocellular carcinoma. Carcinogenesis. 1997;18:445–449. doi: 10.1093/carcin/18.2.445. [DOI] [PubMed] [Google Scholar]
  44. Radack K, Martin V, Wones R, Buncher R, Pinney S, Mandell K. Intercorrelations and sources of variability in three mutagenicity assays: A population-based study. Mutat Res. 1996;350:295–306. doi: 10.1016/0027-5107(95)00162-x. [DOI] [PubMed] [Google Scholar]
  45. Rogatko A, Auerbach AD. Segregation analysis with uncertain ascertainment: Application to Fanconi anemia. Am J Hum Genet. 1988;42:889–897. [PMC free article] [PubMed] [Google Scholar]
  46. Rosendorff J, Bernstein R. Fanconi's anemia—Chromosome breakage studies in homozygotes and heterozygotes. Cancer Genet Cytogenet. 1988;33:175–183. doi: 10.1016/0165-4608(88)90027-1. [DOI] [PubMed] [Google Scholar]
  47. Sala-Trepat M, Boyse J, Richard P, Papadopoulo D, Moustacchi E. Frequencies of HPRT- lymphocytes and glycophorin A variant erythrocytes in Fanconi anemia patients and control donors. Mutat Res. 1993;289:115–126. doi: 10.1016/0027-5107(93)90137-5. [DOI] [PubMed] [Google Scholar]
  48. Schindler D, Seyschab H, Poot M, Hoehn H, Schinzel A, Fryns JP, Tommerup N, Rabinovitch PS. Screening test for ataxia telangiectasia. Lancet. 1987;II:1398–1399. doi: 10.1016/s0140-6736(87)91288-8. [DOI] [PubMed] [Google Scholar]
  49. Schroeder TM, German J. Bloom's syndrome and Fanconi's anemia: Demonstration of two distinctive patterns of chromosome disruption and rearrangement. Humangenetik. 1974;25:299–306. doi: 10.1007/BF00336905. [DOI] [PubMed] [Google Scholar]
  50. Schroeder TM, Anschutz F, Knopp A. Spontane chromosomenaberrationen bei familiaerer panmyelopathie. Humangenetik. 1964;1:194–196. doi: 10.1007/BF00389636. [DOI] [PubMed] [Google Scholar]
  51. Seabright M. A rapid banding technique for human chromosomes. Lancet. 1971;II:971–972. doi: 10.1016/s0140-6736(71)90287-x. [DOI] [PubMed] [Google Scholar]
  52. Worton RG, Grant SG. Segregation-like events in Chinese hamster cells. In: Gottesman MM, editor. Molecular cell genetics. New York: Wiley; 1985. pp. 831–867. [Google Scholar]

RESOURCES