Abstract
Multiple myeloma is the most frequent indication for high-dose melphalan (HDM) chemotherapy with autologous stem cell transplantation (ASCT). Gastrointestinal symptoms represent the most significant non-hematological toxicity of HDM. However, specific, especially genetic, predictors of their incidence or clinical severity are lacking. The amino acid transporters LAT1 and LAT2 encoded by the SLC7A5 and SLC7A8 genes, respectively, are the principal mediators of melphalan uptake into cells. To determine whether genetic variability at these loci contributed to inter-individual differences in the development of gastrointestinal complications of HDM, we analyzed single-nucleotide polymorphisms (SNPs) in these genes in 135 patients with multiple myeloma treated with HDM and ASCT and need for total parenteral nutrition (TPN). Seven SNPs in SLC7A5 and twenty in SLC7A8 were genotyped. Multiple analyses indicated that one SNP in the first intron of SLC7A5, rs4240803, significantly associated with TPN use (OR= 0.45, 95% CI 0.25 – 0.79, p = 0.007). Further, every haplotype that correlated with TPN requirement incorporated this SNP. These results suggest that variability in melphalan transport impacts mucosal injury after HDM. This finding could help in individualizing the dose of this effective and widely used chemotherapeutic agent for multiple myeloma.
Introduction
The alkylating agent melphalan (or phenylalanine mustard) has been a key component in the treatment of multiple myeloma for several decades (1). Its dose-limiting toxicities when used at high doses with autologous stem cell transplantation (ASCT) include myelosuppression and gastrointestinal injury, which can manifest in anorexia, nausea, vomiting, diarrhea, and oral mucositis (OM) [reviewed in (2)].
There is extensive inter-individual variation in melphalan’s pharmacokinetic parameters, both for intravenous and oral administration (3). Further, the frequency and severity of OM, while somewhat related to dose, show considerable variability between patients receiving the identical high dose of melphalan. Genetic differences have been proposed to be responsible (4).
L-type amino acid transporters 1 (LAT1) and 2 (LAT2) are responsible for the transport of large neutral amino acids into the cell. These integral membrane proteins are obligate heterodimers of a catalytic light chain (LAT1 or LAT2) and glycoprotein heavy chain, 4F2hc/CD98 (5). High levels of SLC7A5 mRNA encoding the LAT1 transporter have been noted in most human tissues except for the intestine (6) and LAT1 is thought to play a key role in amino acid uptake by dividing cells. Further, SLC7A5 expression is upregulated in multiple tumor types (7), which may contribute to their abnormal phenotype. In contrast to the broad expression of LAT1, LAT2 is restricted primarily to the kidneys and bowel (8), where it localizes to basolateral membrane and likely functions in epithelial amino acid reabsorption.4
Melphalan resembles phenylalanine in its affinity for LAT1 and LAT2, which mediate the transport of both into cells (9, 10). This study was designed to investigate whether polymorphisms in the SLC7A5 or SLC7A8 genes encoding the catalytic subunits of LAT1 and LAT2, respectively, associate with development of gastrointestinal toxicity after HDM. It revealed a significant association between one SNP, rs4240803, located in the first intron of SLC7A5 and the severity of OM and need for total parenteral nutrition in a series of patients with multiple myeloma from multiple institutions who were all treated with HDM and ASCT.
Materials and Methods
Patients
All subjects carried a diagnosis of multiple myeloma and were treated at either Vanderbilt University Medical Center (n = 114), the Nashville VA Medical Campus of the Tennessee Valley Healthcare System (n = 8), or at the University of Texas Health Sciences Center San Antonio or Audie L. Murphy VA Hospital of San Antonio Veterans Health Care System (n = 13). They could be enrolled prior to or after treatment with HDM chemotherapy and ASCT. This study was approved by the responsible institutional review boards at Vanderbilt University Medical Center, University of Texas Health Science Center at San Antonio, and at the Nashville and San Antonio VA Medical Centers. Informed consent was obtained from all patients before entry. Exclusion criteria included a diagnosis other than multiple myeloma, previous treatment with HDM, and administration of recombinant human keratinocyte or fibroblast growth factor. One hundred and thirty-five Caucasian patients (95 males and 40 females) were included in this analysis. Patient characteristics at entry are listed in Table 1. Peripheral blood samples for DNA extraction were collected at the time of patient enrollment. Patients who received TPN at any time during the first fourteen days after transplant were defined as cases (75 patients, including 43 males and 32 females), while patients who did not receive TPN were treated as controls (60 patients, including 52 males and 8 females).
Table 1.
Patient demographics and melphalan regimen
| Cases | Controls | P-value | |
|---|---|---|---|
| Study Participants | 75 | 60 | |
| Study participants by gender (% female) | 42.67% | 13.33% | 0.000022 |
| Mean age (years) | 57.52 (0.761) | 57.35 (0.981) | 0.79703 |
| Mean melphalan dose (mg/m2) | 196.00 (1.951) | 195.14 (1.921) | 0.75313 |
| Mean body mass index (BMI) | 28.92 (0.531) | 30.23 (0.951) | 0.82313 |
| Pre-treatment chemotherapy (%receiving VAD) | 10.67% | 3.33% | 0.10562 |
Standard error of the mean
P-value presented is from the Two Sample Differences of Proportions test performed in STATA
P-value presented is from the Wilcoxon Rank Sum Test performed in STATA
Treatment
High-dose melphalan was given in two divided doses via an intravenous route three and two days, respectively, prior to transplant. Total doses of 140 mg/m2 (n = 11) or 200 mg/m2 (n = 124) of melphalan were administered. Patients received previously cryopreserved apheresed peripheral blood stem within a range of 2–4 × 106 CD34+ cells/kg of body weight.
Phenotypic Assessment of Gastrointestinal Toxicity
Requirement for total parenteral nutrition (TPN) provided a global measure of gastrointestinal toxicity and served as a proxy for development of mucositis. The indications for TPN use were similar for all centers and included odynophagia requiring narcotic analgesia, severe diarrhea, or inability to eat. Information on the use of TPN after HDM chemotherapy and ASCT was collected on every subject. Use of TPN as an index of gastrointestinal toxicity has the signal advantage of encompassing all of its clinical manifestations, including pain, nausea, and diarrhea, is robust to observer bias, and can be accurately assessed in all patients, either prospectively or retrospectively. For these reasons, TPN requirement was used as the major phenotypic endpoint in our study.
DNA Genotyping and Quality Control
A total of 32 SNPs from the SLC7A5 and SLC7A8 genes were genotyped. These SNPs were selected from their ability to tag nearby variants in the HapMap database (http://www.hapmap.org) using Caucasian samples (CEU). An r2 of 0.80 and a minor allele frequency of 0.10 were used as tagging criteria. Genotyping was performed with the Sequenom MASSarray system in the Vanderbilt University Center for Human Genetics Research Core Laboratory according to the manufacturer’s protocol. Ambiguous calls by the software were resolved by inspection by laboratory personnel and investigators. Of the 32 SNPs genotyped, 27 passed quality control, with a genotyping efficiency of greater than 95%. Seven of the SNPs analyzed were from SLC7A5 and twenty were located in SLC7A8 (Table 2).
Table 2.
Minor allele frequencies of all markers analyzed
| Gene | Marker | Minor Allele | Frequency |
|---|---|---|---|
| SLC7A5 | rs1060253 | G | 0.2889 |
| SLC7A5 | rs3815559 | C | 0.2037 |
| SLC7A5 | rs12931876 | C | 0.1374 |
| SLC7A5 | rs4843718 | A | 0.2519 |
| SLC7A5 | rs731710 | C | 0.3667 |
| SLC7A5 | rs4240803 | A | 0.2815 |
| SLC7A5 | rs17853938 | T | 0.1000 |
| SLC7A8 | rs2331937 | T | 0.2630 |
| SLC7A8 | rs8011016 | C | 0.2333 |
| SLC7A8 | rs4982736 | A | 0.1852 |
| SLC7A8 | rs10150592 | A | 0.1852 |
| SLC7A8 | rs12894506 | T | 0.2293 |
| SLC7A8 | rs6573011 | G | 0.2594 |
| SLC7A8 | rs1983698 | T | 0.1407 |
| SLC7A8 | rs1884545 | T | 0.1111 |
| SLC7A8 | rs910795 | C | 0.1519 |
| SLC7A8 | rs10143650 | T | 0.2841 |
| SLC7A8 | rs10145863 | A | 0.1333 |
| SLC7A8 | rs10132368 | A | 0.1296 |
| SLC7A8 | rs2239628 | G | 0.1481 |
| SLC7A8 | rs7157207 | G | 0.2741 |
| SLC7A8 | rs6572981 | A | 0.2037 |
| SLC7A8 | rs3783436 | C | 0.3545 |
| SLC7A8 | rs4982732 | G | 0.2111 |
| SCL7A8 | rs1015089 | T | 0.3593 |
| SLC7A8 | rs2013931 | C | 0.4037 |
| SLC7A8 | rs999165 | T | 0.2164 |
Statistical Analysis
Prior to analyses for association, all SNPs that could be successfully genotyped were checked for deviations from Hardy-Weinberg equilibrium (HWE) using the Willington’s exact test in PLINK version 1.07 software (11) (http://pngu.mgh.harvard.edu/purcell/plink/) and the results confirmed with the PLATO software package (12) (https://chgr.mc.vanderbilt.edu/plato). Tests were carried out on the entire dataset, as well as on cases and controls independently. Linkage disequilibrium (LD) between markers in SLC7A5 (n = 7) and SLC7A8 (n = 20) was measured using release 4.2 of the software package Haploview (13) (http://haploview@broad.mit.edu). LD was measured in the full dataset, as well as in case and control groups separately (Supplemental Figures 1A–3B). Genotype-phenotype associations were tested using four methods. First, single SNP association was analyzed with Fisher’s exact test of association using PLINK (11), both for the full dataset and a subset containing only male subjects. Females were not analyzed separately owing to their small sample size; however, adjustment for gender was carried out where possible to account for effects. Second, exact logistic regression analysis was used to test for single SNP association with those SNPs shown to be significant in the prior analysis. The results of both analyses are presented. Exact logistic regression analysis was carried out using release 10 of the software package STATA (StataCorp LP, College Station, TX), which also estimated odd ratios. Third, a prevalence-based association test (PRAT) was used to validate logistic regression single SNP analyses (14). PRAT analysis, which is based on Hardy-Weinberg principles, tested the expected case and control genotypes frequency distributions against expected values estimated from the prevalence of phenotypes and allele frequencies within case and control groups. Permutation testing (n = 1000) was used to correct for multiple testing. Finally, two, three, and four SNP window haplotype analysis was performed in both SLC7A5 and SLC7A8 separately using UNPHASED (15) in order to test for multi-locus effects on TPN use. Only common haplotypes with frequencies ≥ 0.05 in either cases or controls were included in our analyses. Odds ratios reported by UNPHASED provided the estimated odds ratios for a given haplotype relative to the reference haplotype, and the most common haplotype was used as the reference in our analyses. Permutation (n = 1000) was used to correct for multiple testing for individual haplotype tests. A figure summarizing study design and the analyses employed is included in Supplemental Materials (Supplemental Figure 4).
Results
Subject Demographics and Clinical Characteristics
Cases and controls were well matched, without statistically significant differences in mean age or melphalan dose (Table 1). They did differ, however, in gender distribution (p = 0.0021), with a significantly higher proportion of females in the group that received TPN (cases) than in the group that did not (controls).
Single Locus Association Analysis
None of the 27 genetic markers that passed quality control deviated significantly from Hardy-Weinberg equilibrium (Supplemental Table 1). Allelic and genotypic associations with TPN use were investigated by Fisher’s exact test. This revealed that of the seven SNPs analyzed in SLC7A5, only one marker, rs4240803, correlated significantly with the use of TPN in both allelic (p = 0.0064) and genotypic (p = 0.0178) analyses (Table 3A). After FDR correction, rs4240803 remained associated with TPN requirement in allelic but not genotypic testing. In contrast, none of twenty markers in SLC7A8 were associated with TPN administration (Table 3A). When analysis was restricted to males only, three SNPs, two in SLC7A5 and one in SLC7A8, were found to have significant allelic and/or genotypic associations with TPN use (Table 3B). The most highly associated in the male-only dataset was rs4240803 (p = 0.0112 for allelic testing and p = 0.0229 for genotypic testing). In addition, two markers in SLC7A8, rs1884545 (p = 0.0212 in genotypic testing) and rs1015089, were associated with TPN requirement (p = 0.0493 in genotypic testing), although less significantly. However, after FDR correction for multiple testing, only SNP rs4242803 still associated with TPN use (for males).
Table 3A.
Genotype distributions and association results for SLC7A5/SLC7A8 with TPN use in the full dataset
| Gene | SNP | Alleles Min/Max | Controls (N)1 | Cases (N)1 | Genotypic p-value | Allelic p-value | ||||
|---|---|---|---|---|---|---|---|---|---|---|
| 11 | 12 | 22 | 11 | 12 | 22 | |||||
| SLC7A5 | rs1060253 | G/C | 5 | 26 | 29 | 4 | 34 | 37 | 0.8270 | 0.7873 |
| rs3815559 | C/G | 3 | 20 | 37 | 4 | 21 | 50 | 0.8706 | 0.6511 | |
| rs12931876 | C/T | 2 | 14 | 43 | 2 | 14 | 56 | 0.8759 | 0.5897 | |
| rs4843718 | A/G | 5 | 23 | 32 | 7 | 21 | 47 | 0.4686 | 0.4814 | |
| rs731710 | C/T | 11 | 28 | 21 | 9 | 31 | 35 | 0.3371 | 0.1301 | |
| rs4240803 | A/G | 7 | 30 | 23 | 3 | 26 | 46 | 0.0178 | 0.0064 | |
| rs17853938 | T/G | 1 | 10 | 49 | 0 | 15 | 60 | 0.6468 | 1.0000 | |
| SLC7A8 | rs2239628 | G/A | 0 | 15 | 45 | 3 | 19 | 53 | 0.3988 | 0.3907 |
| rs7157207 | G/T | 3 | 27 | 30 | 6 | 29 | 40 | 0.6531 | 1.0000 | |
| rs6572981 | A/G | 0 | 23 | 37 | 4 | 24 | 47 | 0.1856 | 0.7613 | |
| rs3783436 | C/T | 12 | 22 | 26 | 8 | 33 | 33 | 0.3252 | 0.4412 | |
| rs4982732 | G/C | 2 | 19 | 39 | 3 | 28 | 44 | 0.7774 | 0.5493 | |
| rs1015089 | T/G | 4 | 32 | 24 | 13 | 31 | 31 | 0.1283 | 0.4463 | |
| rs2013931 | C/T | 12 | 24 | 24 | 13 | 35 | 27 | 0.7632 | 1.0000 | |
| rs999165 | T/A | 3 | 21 | 36 | 3 | 25 | 46 | 0.9553 | 0.7676 | |
| rs2331937 | T/C | 3 | 23 | 34 | 7 | 28 | 40 | 0.7123 | 0.4905 | |
| rs8011016 | C/T | 5 | 17 | 38 | 2 | 32 | 41 | 0.1171 | 0.8850 | |
| rs4982736 | A/G | 3 | 20 | 37 | 3 | 18 | 54 | 0.4517 | 0.2706 | |
| rs10150592 | A/C | 3 | 19 | 38 | 3 | 19 | 53 | 0.6683 | 0.4316 | |
| rs910795 | C/T | 2 | 16 | 42 | 3 | 15 | 57 | 0.7092 | 0.6098 | |
| rs10143650 | T/C | 4 | 20 | 36 | 5 | 29 | 41 | 0.8556 | 0.6716 | |
| rs10145863 | A/G | 1 | 13 | 46 | 0 | 21 | 54 | 0.4218 | 0.8573 | |
| rs12894506 | T/C | 2 | 20 | 37 | 4 | 29 | 41 | 0.7152 | 0.3829 | |
| rs6573011 | G/A | 3 | 21 | 36 | 5 | 32 | 36 | 0.4790 | 0.2635 | |
| rs10132368 | A/G | 1 | 10 | 49 | 3 | 17 | 55 | 0.5453 | 0.2078 | |
| rs1983698 | T/C | 0 | 18 | 42 | 1 | 18 | 56 | 0.6851 | 0.7271 | |
| rs1884545 | T/C | 0 | 16 | 44 | 2 | 10 | 63 | 0.0620 | 0.3333 | |
P values in bold remained significant after correction for multiple testing.
Genotype key – 11: homozygous minor; 12: heterozygous; 22: homozygous major. P-values showns are Fisher’s exact p-values.
Table 3B.
Genotype distributions and association results for SLC7A5/SLC7A8 with TPN use in males only.
| Gene | SNP | Alleles Min/Max | Controls (N)1 | Cases (N)1 | Genotypic p-value | Allelic p-value | ||||
|---|---|---|---|---|---|---|---|---|---|---|
| 11 | 12 | 22 | 11 | 12 | 22 | |||||
| SLC7A5 | rs1060253 | G/C | 4 | 23 | 25 | 2 | 18 | 23 | 0.8489 | 0.6262 |
| rs3815559 | C/G | 2 | 18 | 32 | 3 | 10 | 30 | 0.4325 | 0.7178 | |
| rs12931876 | C/T | 1 | 13 | 37 | 1 | 6 | 35 | 0.4364 | 0.372 | |
| rs4843718 | A/G | 4 | 20 | 28 | 5 | 9 | 29 | 0.1764 | 0.5010 | |
| rs731710 | C/T | 10 | 24 | 18 | 6 | 15 | 22 | 0.2762 | 0.1341 | |
| rs4240803 | A/G | 6 | 28 | 18 | 2 | 14 | 27 | 0.0229 | 0.0112 | |
| rs17853938 | T/G | 0 | 10 | 42 | 0 | 7 | 36 | 0.7920 | 0.8022 | |
| SLC7A8 | rs2239628 | G/A | 0 | 14 | 38 | 3 | 10 | 30 | 0.1985 | 0.4245 |
| rs7157207 | G/T | 3 | 25 | 24 | 5 | 17 | 21 | 0.5303 | 0.8747 | |
| rs6572981 | A/G | 0 | 21 | 31 | 3 | 14 | 26 | 0.1406 | 0.7234 | |
| rs3783436 | C/T | 10 | 21 | 21 | 2 | 19 | 22 | 0.1064 | 0.0895 | |
| rs4982732 | G/C | 2 | 15 | 35 | 2 | 18 | 23 | 0.3558 | 0.2878 | |
| rs1015089 | T/G | 3 | 27 | 22 | 10 | 19 | 14 | 0.0493 | 0.0712 | |
| rs2013931 | C/T | 10 | 22 | 20 | 4 | 22 | 17 | 0.3937 | 0.4563 | |
| rs999165 | T/A | 0 | 19 | 33 | 1 | 13 | 29 | 0.5100 | 1.0000 | |
| rs2331937 | T/C | 3 | 20 | 29 | 6 | 15 | 22 | 0.4433 | 0.3353 | |
| rs8011016 | C/T | 5 | 14 | 33 | 2 | 18 | 23 | 0.7969 | 0.8631 | |
| rs4982736 | A/G | 3 | 19 | 30 | 3 | 13 | 27 | 0.7969 | 0.8631 | |
| rs10150592 | A/C | 3 | 18 | 31 | 3 | 14 | 26 | 1.0000 | 1.0000 | |
| rs910795 | C/T | 2 | 15 | 35 | 3 | 12 | 28 | 0.8252 | 0.7141 | |
| rs10143650 | T/C | 4 | 18 | 30 | 4 | 21 | 18 | 0.2988 | 0.2018 | |
| rs10145863 | A/G | 1 | 12 | 39 | 0 | 13 | 30 | 0.6377 | 0.8355 | |
| rs12894506 | T/C | 2 | 18 | 31 | 3 | 22 | 18 | 0.1879 | 0.0996 | |
| rs6573011 | G/A | 3 | 20 | 29 | 2 | 23 | 17 | 0.2965 | 0.3287 | |
| rs10132368 | A/G | 1 | 10 | 41 | 1 | 13 | 29 | 0.4615 | 0.2983 | |
| rs1983698 | T/C | 0 | 17 | 35 | 1 | 8 | 34 | 0.1612 | 0.4079 | |
| rs1884545 | T/C | 0 | 15 | 37 | 1 | 4 | 38 | 0.0212 | 0.1619 | |
P values in bold remained significant after correction for multiple testing.
Genotype key-11: homozygous minor; 12: heterozygous, 22: homozygous major. P-values shown are Fisher’s exact p-values.
Univariate exact logistic regression analysis was carried out for all SNPs found to be significantly associated with clinical phenotype in Fisher’s exact analysis, both for the complete dataset and for males only (Table 4). The reference genotype used in this analysis was the homozygote for the major allele. Analysis of the full dataset showed that rs4240803 was significantly associated with TPN use (OR = 0.45; 95% CI = 0.25 – 0.79; p = 0.007). Of the three SNPs associated with TPN requirement in males, rs4240803, rs1015089, and rs1884545 (in SLC7A8), only rs4240803 was significantly associated in exact logistic regression analysis (OR = 0.40; 95% CI = 0.19 – 0.80; p = 0.011). The most significantly associating SNP in the complete dataset, rs4240803, showed slightly lower, although still significant, association when analysis was restricted to males for either analytical method, suggesting that the genetic effect trended in the same direction for both genders and that significance was likely decreased by the reduction in sample size after filtering out female participants. However, the discovery of markers in SLC7A8 that associated with TPN in the male-only subset but not the full dataset suggest that the genetic effects of these markers could, in principle, differ by gender; however, such a conclusion is at present unwarranted.
Table 4.
Logistic regression results for full and male-only datasets.
| SNP | Dataset Analyzed1 | Odds Ratio | 95% Confidence Interval | p-value |
|---|---|---|---|---|
| SLC7A5 | ||||
| rs4240803 | FULL | 0.45 | 0.25–0.79 | 0.007 |
| MALE ONLY | 0.40 | 0.19–0.80 | 0.011 | |
| SLC7A8 | ||||
| rs1015089 | MALE ONLY | 1.83 | 0.95–3.63 | 0.072 |
| rs1884545 | MALE ONLY | 0.44 | 0.13–1.29 | 0.158 |
This denotes the dataset in which each logistic regression analysis was performed.
Association between the SNPs in SLC7A5 and SLC7A8 and TPN requirement was also analyzed with PRAT (Tables 5A and 5B), which serves to validate associations detected using logistic regression analysis. PRAT analysis of the full dataset identified three SNPs associated with TPN use (Table 5A). Two of the markers were discovered in previous analyses, rs4240803 and rs1015089, while a third, novel association was uncovered for rs6572981 in SLC7A8. The single SLC7A5 SNP rs4240803 showed a significant association only for cases (p = 0.0070) and was the most significantly associated SNP in PRAT analysis of the full dataset. Rs1015089 and rs6572981 were marginally significant in controls only, with p = 0.043 and 0.044, respectively. PRAT analysis in the male-only subset (Table 5B) confirmed the significant association between rs4240803 and TPN use (p = 0.011 in cases), while rs6572981 was marginally significant (p = 0.049). In contrast to the results with the full dataset, rs1015089 became more significant in controls (p = 0.029) and displayed only marginal significance in cases (p = 0.046). In addition, rs1884545, which was not identified in previous PRAT analysis of the full dataset, was found to be associated with TPN requirement in PRAT analysis of males (p = 0.012).
Table 5A.
Prevalence-based Assocation Test (PRAT) results in full dataset
| Gene | SNP | PRAT p-value (cases)1 | PRAT p-value (controls)2 |
|---|---|---|---|
| SLC7A5 | rs1060253 | 0.5800 | 0.9520 |
| rs3815559 | 0.6370 | 1.0000 | |
| rs12931876 | 0.6710 | 0.6770 | |
| rs4843718 | 0.3370 | 0.8840 | |
| rs731710 | 0.2230 | 0.6310 | |
| rs4240803 | 0.0070 | 0.2210 | |
| rs17853938 | 0.6740 | 0.7640 | |
| SLC7A8 | rs2239628 | 0.3020 | 0.2010 |
| rs7157207 | 1.0000 | 0.2750 | |
| rs6572981 | 0.8420 | 0.0440 | |
| rs3783436 | 0.7430 | 0.0860 | |
| rs4982732 | 0.6230 | 0.9170 | |
| rs1015089 | 0.3060 | 0.0430 | |
| rs2013931 | 0.9890 | 0.2400 | |
| rs999165 | 0.9110 | 1.0000 | |
| rs2331937 | 0.5280 | 0.7450 | |
| rs8011016 | 0.2320 | 0.0720 | |
| rs4982736 | 0.3780 | 0.8180 | |
| rs10150592 | 0.5590 | 0.8090 | |
| rs910795 | 0.5300 | 0.8970 | |
| rs10143650 | 0.8040 | 0.6230 | |
| rs10145863 | 0.2840 | 0.9380 | |
| rs12894506 | 0.5480 | 0.7090 | |
| rs6573011 | 0.3340 | 0.7510 | |
| rs10132368 | 0.2170 | 0.7090 | |
| rs1983698 | 0.9120 | 0.1520 | |
| rs1884545 | 0.1250 | 0.1100 |
P values in bold remained significant after correction for multiple testing.
PRAT p-values in cases-only based on 1000 permutations.
PRAT p-values in controls only based on 1000 permutations.
Table 5B.
Prevalence-based Association Test (PRAT) results in males-only subset
| Gene | SNP | PRAT p-value (cases)1 | PRAT p-value (controls)2 |
|---|---|---|---|
| SLC7A5 | rs1060253 | 0.6580 | 0.7080 |
| rs3815559 | 0.3740 | 0.9250 | |
| rs12931876 | 0.3610 | 0.7710 | |
| rs4843718 | 0.2050 | 0.9690 | |
| rs731710 | 0.1930 | 0.6900 | |
| rs4240803 | 0.0110 | 0.0970 | |
| rs17853938 | 1.0000 | 0.4940 | |
| SLC7A8 | rs2239628 | 0.1090 | 0.2210 |
| rs7157207 | 0.8400 | 0.2080 | |
| rs6572981 | 0.6800 | 0.0490 | |
| rs3783436 | 0.1520 | 0.0940 | |
| rs4982732 | 0.3080 | 0.6190 | |
| rs1015089 | 0.0460 | 0.0290 | |
| rs2013931 | 0.5210 | 0.2530 | |
| rs999165 | 1.0000 | 0.1830 | |
| rs2331937 | 0.2360 | 0.8490 | |
| rs8011016 | 0.7140 | 0.0710 | |
| rs4982736 | 0.6870 | 1.0000 | |
| rs10150592 | 0.8580 | 0.9460 | |
| rs910795 | 0.6090 | 0.9240 | |
| rs10143650 | 0.2740 | 0.3980 | |
| rs10145863 | 0.3490 | 0.8930 | |
| rs12894506 | 0.0670 | 0.5720 | |
| rs6573011 | 0.1350 | 0.8910 | |
| rs10132368 | 0.4830 | 0.6360 | |
| rs1983698 | 0.5190 | 0.0590 | |
| rs1884545 | 0.0590 | 0.0120 |
P values in bold remained significant after correction for multiple testing.
PRAT p-values in cases-only based on 1000 permutations.
PRAT p-values in controls-only based on 1000 permutations.
Haplotype Analysis
Haplotype association testing with two to four SNP sliding windows was performed on the full dataset using UNPHASED. Significant haplotype associations were discovered only for SLC7A5 (Figure 1, Table 6A–6C). The haplotype most highly associated with TPN use was a four-locus haplotype group comprised of rs4843718, rs731710, rs4240803, and rs17853938 (C-C-A-G). The association with TPN requirement (p = 0.003) remained significant even after correction for multiple testing. Importantly, each of the six haplotypes that associated with TPN use included rs4240803, indicating that all of the haplotype effects observed were driven by this SNP.
Table 6A.
Signifcant 2 locus haplotype associations in the SLC7A5 gene with use of TPN
| SNPs | Haplotype | Frequency Cases | Frequency Controls | Odds Ratio | 95% Confidence Interval | P-value1 |
|---|---|---|---|---|---|---|
| rs731710-rs4240803 | 0.03442 | |||||
| Reference Haplotype | C-A | 0.1304 | 0.5551 | 1.00 | 1.00–1.00 | 0.0048 |
| C-G | 0.1963 | 0.0496 | 2.72 | 1.10–6.72 | 0.3731 | |
| T-A | 0.083 | 0.3382 | 1.73 | 0.60–5.37 | 0.4481 | |
| T-G | 0.5094 | 0.0541 | 2.51* | 1.28–4.89 | 0.0540 | |
| rs4240803-rs17853938 | 0.02872 | |||||
| Reference Haplotype | A-G | 0.2133 | 0.3507 | 1.00 | 1.00–1.00 | 0.0087 |
| A-T | 0.0000 | 0.0160 | - | - | - | |
| G-G | 0.6867 | 0.5493 | 2.06* | 1.17–3.60 | 0.0157 | |
| G-T | 0.1000 | 0.0840 | 1.96 | 0.77–4.99 | 0.7472 | |
P values in bold remained significant after correction for multiple testing.
-Rare haplotype (frequency <0.05 in either cases or controls) excluseded from individual haplotype analysis.
Odds ratio is statistically significant.
P-value for each individual haplotype vs. all others based on UNPHASED score statistic.
Global p-value for a particular group of loci.I is derived from a likelihood ratio test comparing the overall distribution of haplotypes between cases and controls and tests hypothesis that none of the haplotypes for a given set of loci are associated with TPN use.
Table 6C.
Significant 4 locus haplotype associations in the SLC7A5 gene with use of TPN
| SNPs | Haplotype | Frequency Cases | Frequency Controls | Odds Ratio | 95% Confidence Interval | P-value1 |
|---|---|---|---|---|---|---|
| rs12931876-rs4843718-rs731710-rs4240803 | 0.02322 | |||||
| Reference Haplotype | G-T-T-G | 0.5878 | 0.4838 | 1.00 | 1.00–1.00 | 0.0673 |
| C-C-C-A | 0.0269 | 0.1294 | 0.17* | 0.05–0.62 | 0.0036 | |
| C-C-C-G | 0.0832 | 0.0231 | 2.97 | 0.63–13.95 | 0.10.36 | |
| C-T-C-A | 0.0204 | 0.0313 | - | - | - | |
| C-T-C-G | 0.0432 | 0.0280 | - | - | - | |
| G-C-T-G | 0.0149 | 0.0000 | - | - | - | |
| G-T-C-A | 0.0832 | 0.1027 | 0.67 | 0.27–1.64 | 0.3718 | |
| G-T-C-G | 0.0696 | 0.1007 | 0.57 | 0.22–1.50 | 0.5288 | |
| G-T-T-A | 0.0709 | 0.1009 | 0.58 | 0.21–1.61 | 0.2581 | |
| rs4843718-rs731710-rs4240803-rs17853938 | 0.02482 | |||||
| Reference Haplotype | T-T-G-G | 0.518 | 0.4404 | 1.00 | 1.00–1.00 | 0.1223 |
| C-C-A-G | 0.0240 | 0.1299 | 0.16* | 0.04–0.64 | 0.0031 | |
| C-C-G-G | 0.0822 | 0.0227 | 3.08 | 0.65–14.57 | 0.0942 | |
| C-T-G-G | 0.0188 | 0.0000 | - | - | - | |
| T-C-A-G | 0.1064 | 0.1172 | 0.77 | 0.33–1.81 | 0.4429 | |
| T-C-A-T | 0.0000 | 0.0164 | - | - | - | |
| T-C-G-G | 0.0747 | 0.0873 | 0.73 | 0.26–2.07 | 0.9256 | |
| T-C-G-T | 0.0391 | 0.0418 | - | - | - | |
| T-T-A-G | 0.0710 | 0.1009 | 0.60 | 0.21–1.67 | 0.2630 | |
| T-T-G-T | 0.0650 | 0.0435 | 1.27 | 0.36–4.47 | 0.5210 | |
P values in bold remained significant after correction for multiple testing.
-Rare haplotype (frequency <0.05 in either cases or controls) excluseded from individual haplotype analysis.
Odds ratio is statistically significant.
P-value for each individual haplotype vs. all others based on UNPHASED score statistic.
Global p-value for a particular group of loci.I is derived from a likelihood ratio test comparing the overall distribution of haplotypes between cases and controls and tests hypothesis that none of the haplotypes for a given set of loci are associated with TPN use.
Discussion
After myelosuppression, gastrointestinal injury is the most frequent toxicity of HDM and ASCT. The development of mucositis in this setting translates into an increased risk of infection, requirement for parenteral nutrition and intravenous narcotics, greater expense, longer hospitalization, and, possibly, higher mortality (16).
Our study identified a SNP in the first intron of LAT1, rs4240803, which was associated with a requirement for TPN in patients who received HDM and ASCT for multiple myeloma. The significance of this SNP was demonstrated in both allelic and genotypic analysis and remained significant after FDR correction in allelic testing. Independently, haplotype association analysis indicated that multi-locus effects were largely due to rs4240803. The association between this SNP and TPN use, which was confirmed in exact logistic regression analysis and validated with PRAT analysis, was also significant in male subjects only, suggesting that the direction of the genetic effect was the same for both genders. Interestingly, there was another marker in our analyses, rs1884545 (in SLC7A8), for which a significant association was identified in the maleonly subset. However, females were under-represented in our study population as a result of two of the four study locations being situated in VA medical centers, where the majority of patients are male, thereby precluding any definitive conclusions about gene by gender effects.
TPN requirement was utilized as the primary clinical endpoint, both because it encompasses injury arising anywhere within the gastrointestinal tract and because it is well suited to retrospective or prospective analysis of toxicity. Additionally, the lack of availability of formal mucositis scores for the majority of our study population and the susceptibility of this type of measurement to observer bias, a real consideration since four transplant units affiliated with two different medical centers were utilized in this study, also served to reduce the value of utilizing, for example, the Oral Mucositis Assessment (OMAS) or World Health Organization scales for phenotypic analysis. However, we acknowledge that these scales have been experimentally validated and therefore carried out an exploratory analysis of patients where such data had been collected to assess whether OMAS scores, and, by proxy, severity of mucositis was influenced by rs4240803. This analysis, which could be carried out for only a minority of subjects, revealed that maximum OMAS score during treatment was significantly associated with rs4240803 genotype (p = 0.003) (Supplemental Material). Finally, as LAT1 protein is expressed on endothelial cells that contribute to the blood-brain barrier, it is possible that melphalan transport into the central nervous system correlated with transport into epithelial cells in the gut, contributing to nausea and anorexia and via this mechanism also impacting TPN use.
These results, at least indirectly, underscore the importance of LAT1 as the major transporter of melphalan into cells, and localization of SNP rs4240803 within the first intron of the SLC7A5 gene close by a region bearing an epigenetic signature of a transcriptional enhancer (data not shown) has direct implications for mechanism. Either by contributing to or, more likely, linked in cis to a polymorphism affecting the binding of a transcription factor to this putative enhancer, genetic variation could impact SLC7A5 transcription, membrane expression of LAT1, and ultimately melphalan transport into gastrointestinal epithelial cells. Alternatively, the polymorphism could be in linkage disequilibrium with a variant that affects drug pharmacokinetics, and although LAT1-mediated cellular uptake could affect drug handling, it should be noted that association between drug pharmacokinetics and toxicity has not been observed (3).
While our work was in progress, a similarly designed study was published that failed to detect an association between SNPs in the genes encoding LAT1, LAT2, and 4F2hc and gastrointestinal toxicity following HDM chemotherapy (17). Although the patient population used was more heterogeneous than the one in our study and included individuals with diseases other than multiple myeloma, the more likely reason for these discrepant results was the lack of representation in their analysis of SNPs from the very large first intron of SLC7A5, such as rs4240803.
An important implication of this work is the existence of pharmacogenetic determinants of toxicity of HDM, which could prove valuable in individualizing drug dose. This study involved both prospectively and retrospectively acquired data, and the association between SNP rs4240803, and possibly several other markers in SLC7A5 or SLC7A8 and gastrointestinal toxicity after HDM, measured by TPN use and/or mucositis grade, requires validation in another patient population from which data are acquired in a wholly prospective fashion. Such a study is underway, together with a systematic analysis of SLC7A5 intronic sequences for enhancer function.
Supplementary Material
Table 6B.
Significant 3 locus haplotype associations in the SLC7A5 gene with use of TPN
| SNPs | Haplotype | Frequency Cases | Frequency Controls | Odds Ratio | 95% Confidence Interval | P-value1 |
|---|---|---|---|---|---|---|
| rs4843718 – rs731710 – rs4240803 | 0.00992 | |||||
| Reference Haplotype | T-T-G | 0.5845 | 0.4837 | 1.00 | 1.00–1.00 | 0.0687 |
| C-C-G | 0.0815 | 0.0230 | 2.93 | 0.62–13.81 | 0.1084 | |
| C-T-G | 0.0183 | 0.0000 | - | - | - | |
| T-C-A | 0.1055 | 0.1338 | 0.65 | 0.29–1.47 | 0.2468 | |
| T-C-G | 0.1143 | 0.1289 | 0.73 | 0.32–1.70 | 0.9366 | |
| T-T-A | 0.0708 | 0.1011 | 0.58 | 0.21–1.61 | 0.2651 | |
| C-C-A | 0.0252 | 0.1295 | 0.16* | 0.04–0.62 | 0.0034 | |
| rs731710-rs4240803-rs17853938 | 0.11022 | |||||
| Reference Haplotype | T-G-G | 0.5285 | 0.4346 | 1.00 | 1.00–1.00 | 00.755 |
| C-A-G | 0.1312 | 0.2670 | 0.40* | 0.20–0.80 | 0.0055 | |
| C-G-G | 0.1582 | 0.1086 | 1.20 | 0.51–2.80 | 0.3117 | |
| C-G-T | 0.0373 | 0.0411 | - | - | - | |
| T-A-G | 0.0822 | 0.899 | 0.75 | 0.28–2.03 | 0.5897 | |
| T-A-T | 0.0000 | 0.0098 | - | - | - | |
| T-G-T | 0.0627 | 0.0491 | 1.05 | 0.30–3.65 | 0.6799 | |
P values in bold remained significant after correction for multiple testing.
-Rare haplotype (frequency <0.05 in either cases or controls) excluseded from individual haplotype analysis.
Odds ratio is statistically significant.
P-value for each individual haplotype vs. all others based on UNPHASED score statistic.
Global p-value for a particular group of loci.I is derived from a likelihood ratio test comparing the overall distribution of haplotypes between cases and controls and tests hypothesis that none of the haplotypes for a given set of loci are associated with TPN use.
Acknowledgments
This work was supported in part by CTSA award UL1TR000445 from the National Center for Advancing Translational Sciences, National Institutes of Health.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
References
- 1.Rajkumar SV, Kyle RA. Multiple myeloma: diagnosis and treatment. Mayo Clinic Proceedings. 2005;80:1371–1382. doi: 10.4065/80.10.1371. [DOI] [PubMed] [Google Scholar]
- 2.Bayraktar UD, Bashir Q, Qazilbash M, Champlin RE, Ciurea SO. Fifty years of melphalan use in hematopoietic stem cell transplantation. Biol Blood Marrow Transplant. 2013;19:344–356. doi: 10.1016/j.bbmt.2012.08.011. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Mougenot P, Pinguet F, Fabbro M, et al. Population pharmacokinetics of melphalan, infused over a 24-hour period, in patients with advanced malignancies. Cancer Chemother Pharmacol. 2004;53:503–512. doi: 10.1007/s00280-003-0761-2. [DOI] [PubMed] [Google Scholar]
- 4.Grazziutti ML, Dong L, Miceli MH, et al. Oral mucositis in myeloma patients undergoing melphalan-based autologous stem cell transplantation: incidence, risk factors and a severity predictive model. Bone Marrow Transplant. 2006;38:501–506. doi: 10.1038/sj.bmt.1705471. [DOI] [PubMed] [Google Scholar]
- 5.Verrey F. System L: heteromeric exchangers of large, neutral amino acids involved in directional transport. Pflugers Arch. 2003;445:529–533. doi: 10.1007/s00424-002-0973-z. [DOI] [PubMed] [Google Scholar]
- 6.Prasad PD, Wang H, Huang W, et al. Human LAT1, a subunit of system L amino acid transporter: molecular cloning and transport function. Biochem Biophys Res Commun. 1999;255:283–288. doi: 10.1006/bbrc.1999.0206. [DOI] [PubMed] [Google Scholar]
- 7.Yanagida O, Kanai Y, Chairoungdua A, et al. Human L-type amino acid transporter 1 (LAT1): characterization of function and expression in tumor cell lines. Biochim Biophys Acta. 2001;1514:291–302. doi: 10.1016/s0005-2736(01)00384-4. [DOI] [PubMed] [Google Scholar]
- 8.Rossier G, Meier C, Bauch C, et al. LAT2, a new basolateral 4F2hc/CD98-associated amino acid transporter of kidney and intestine. J Biol Chem. 1999;274:34948–34954. doi: 10.1074/jbc.274.49.34948. 17. [DOI] [PubMed] [Google Scholar]
- 9.del Amo EM, Urtti A, Yliperttula M. Pharmacokinetic role of L-type amino acid transporters LAT1 and LAT2. Eur J Pharm Sci. 2008;35:161–174. doi: 10.1016/j.ejps.2008.06.015. [DOI] [PubMed] [Google Scholar]
- 10.Uchino H, Kanai Y, Kim DK, et al. Transport of amino acid-related compounds mediated by L-type amino acid transporter 1 (LAT1): insights into the mechanisms of substrate recognition. Mol Pharmacol. 2002;61:729–737. doi: 10.1124/mol.61.4.729. [DOI] [PubMed] [Google Scholar]
- 11.Purcell S, Neale B, Todd-Brown K, et al. PLINK: a tool set for whole-genome association and population-based linkage analyses. Am J Hum Genet. 2007;81:559–575. doi: 10.1086/519795. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Grady BJ, Torstenson E, Dudek SM, Giles J, Sexton D, Ritchie MD. Finding unique filter sets in plato: a precursor to efficient interaction analysis in gwas data. Pac Symp Biocomput. 2010:315–326. [PMC free article] [PubMed] [Google Scholar]
- 13.Barrett JC, Fry B, Maller J, Daly MJ. Haploview: analysis and visualization of LD and haplotype maps. Bioinformatics. 2005;21:263–265. doi: 10.1093/bioinformatics/bth457. [DOI] [PubMed] [Google Scholar]
- 14.Ryckman KK, Jiang L, Li C, Bartlett J, Haines JL, Williams SM. A prevalence-based association test for case-control studies. Genet Epidemiol. 2008;32:600–605. doi: 10.1002/gepi.20342. [DOI] [PubMed] [Google Scholar]
- 15.Dudbridge F. Pedigree disequilibrium tests for multilocus haplotypes. Genet Epidemiol. 2003;25:115–121. doi: 10.1002/gepi.10252. [DOI] [PubMed] [Google Scholar]
- 16.Sonis ST, Oster G, Fuchs H, et al. Oral mucositis and the clinical and economic outcomes of hematopoietic stem-cell transplantation. J Clin Oncol. 2001;19:2201–2205. doi: 10.1200/JCO.2001.19.8.2201. [DOI] [PubMed] [Google Scholar]
- 17.Kuhne A, Kaiser R, Schirmer M, et al. Genetic polymorphisms in the amino acid transporters LAT1 and LAT2 in relation to the pharmacokinetics and side effects of melphalan. Pharmacogenet Genomics. 2007;17:505–517. doi: 10.1097/FPC.0b013e3280ea77cd. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.