POS0393 ASSOCIATION BETWEEN GENETICALLY PREDICTED EXPRESSION OF TPMT AND AZATHIOPRINE ADVERSE EVENTS IN PATIENTS WITH INFLAMMATORY CONDITIONS

Background

Azathioprine is a widely used immunosuppressant for the treatment of inflammatory conditions such as systemic lupus erythematosus (SLE), systemic vasculitis, dermatomyositis, and inflammatory bowel disease. However, its use is often limited by myelotoxicity. Variants in the gene encoding thiopurine-S-methyltransferase (TPMT), an enzyme in the metabolic pathway of azathioprine, increase the risk for myelotoxicity1. We know little about the relationship between the genetically predicted expression of TPMT and side effects of azathioprine.

Objectives

To examine whether genetically predicted expression of TPMT in liver tissue is associated with azathioprine adverse effects.

Methods

We assembled a retrospective cohort of new users taking azathioprine for inflammatory conditions at a tertiary care center. We performed genotyping with Illumina Infinium Expanded Multi-Ethnic Genotyping Array plus custom content data, and we then used Michigan Imputation servers for genetic imputation and PrediXcan models trained with GTEx/Genotype-Tissue Expression Project version 8 data to impute TPMT expression in liver tissue. We prespecified nine groups of phecodes (comprised of ICD9 and ICD10 codes) corresponding to known adverse effects of azathioprine. We then tested the association between the predicted expression of TPMT and these adverse events; for outcomes significant in the Wilcoxon ranksum tests (p<0.05), each case was reviewed in clinical records for confirmation. Finally, we grouped the predicted expression of TPMT in liver tissue into tertiles and conducted logistic regressions to assess the associations between predicted expression and side effects. We conducted a sensitivity analysis restricted to patients with EHR-reported White race.

Results

The cohort included 1034 patients (Table 1). Phecodes for 3 side effects—leukopenia (n=29), skin cancer (n=13), and rash (n=52)—were identified as associated with predicted TPMT expression in liver tissue. Of these, cases of side effects attributed to azathioprine were validated by chart review: leukopenia (96.6%; n=28), skin cancer (92.3%; n=12), and rash (9.6%; n=5) and used for analysis. When assessed by tertile of predicted TPMT expression, patients in the highest tertile had lower odds of having leukopenia (OR=0.35, 95%CI: 0.12-0.98, p=0.045) and a trend towards higher odds for skin cancer, but the number of cases was small (OR=3.56, 95%CI: 0.73-17.27, p=0.115). Confirmed cases of rash attributed to azathioprine were too few for meaningful analysis. We found similar results when restricted to patients with reported White race.

Table 1.

Characteristics of patients by TPMT expression

Lowest TertileMiddle TertileHighest TertileN=345N=345N=344Female sex, n (%)228 (66.1)244 (70.7)238 (69.2)EHR-reported White race, n (%)306 (88.7)293 (84.9)290 (84.3)Age, median [IQR]42 [29-58]43 [30-55]46 [30-56]Indication, n (%)Systemic lupus erythematosus38 (11.0)42 (12.2)38 (11.0)Inflammatory bowel disease191 (55.3)185 (53.6)190 (55.2)Other connective tissue disorder/autoimmune92 (26.7)96 (27.8)100 (29.1)Other24 (7.0)22 (6.4)16 (4.7)Verified leukopenia attributed to azathioprine, n (%)14 (4.1)9 (2.6)5 (1.5)Verified skin cancer attributed to azathioprine, n (%)2 (0.6)3 (0.9)7 (2.0)Verified rash attributed to azathioprine, n (%)2 (0.6)3 (0.9)0 (0)

Conclusion

This analysis suggests that PrediXcan may be useful for examining the association between gene expression and side effects of medications. Moreover, this approach successfully identified leukopenia as a side effect associated with predicted TPMT expression.

Acknowledgements

None to declare.

Disclosure of Interests

None declared.

POS1444 FLT1 AND EPHB2 ARE NOVEL GENETIC MARKERS ASSOCIATED WITH PANCREATITIS IN PATIENTS TAKING AZATHIOPRINE FOR IMMUNE-MEDIATED CONDITIONS: INTEGRATING GENOME- AND TRANSCRIPTOME-WIDE ASSOCIATION STUDIES

Background

Azathioprine (AZA) is a thiopurine immunosuppressant medication used to treat a variety of immune-mediated diseases. Unfortunately, its use is limited by adverse effects. Pancreatitis, a potentially severe, life-threatening side effect is independent of dose and necessitates AZA discontinuation given the high risk of recurrent pancreatitis with continued use or re-challenge. The mechanisms driving pancreatitis are unclear. While classic thiopurine-induced acute pancreatitis (TIAP) has been associated with HLA haplotypes, most patients taking AZA and presenting with pancreatitis do not fulfill the stringent criteria for TIAP.

Objectives

To identify genetic risk factors for pancreatitis in patients taking azathioprine for immune-mediated conditions.

Methods

Using a biobank linked to electronic health records (EHR) from a tertiary center, we identified new users of AZA. Patients were excluded if the primary indication for AZA was organ transplant or if there was a history of pancreatitis prior to AZA use. The analysis was restricted to patients with EHR-reported race as White due to insufficient case counts for the non-White group. We then identified patients with amylase or lipase values that exceeded twice the upper limit of normal (“>2x ULN”) or with ICD-9/ICD-10 codes for acute pancreatitis. Each record was manually reviewed to confirm the timing of AZA use in relation to laboratory derangements or ICD coding, as well as to further classify patients into three increasingly strict, but not exclusive categories: 1) pancreatic injury (amylase or lipase >2x ULN); 2) acute pancreatitis1, or 3) TIAP2. We completed genotyping with Illumina Infinium Expanded Multi-Ethnic Genotyping Array plus custom content data, employed Michigan Imputation servers for genetic imputation, and used PrediXcan (GTEx v8) to impute gene expression. We then conducted genome-wide association and transcriptome-wide association studies (GWAS, TWAS). Acknowledging the relatively small overall cohort, and possible imbalance of cases vs controls, we used the Firth logistic regression method, which is a penalized likelihood-based method.

Results

We studied 2127 AZA users (35.4% male; mean 44.5+/-17.2 years). The median AZA dose was 100mg/day (IQR: 50-125mg/day). Rheumatologic conditions (56.9%) and inflammatory bowel disease (40.4%) comprised the most common primary indications for AZA. Pancreatic injury, pancreatitis, and TIAP were diagnosed in 42 (2.0%), 16 (0.8%), and 9 (0.4%) patients, respectively. GWAS identified several significantly associated genes, many with overlapping TWAS findings in the pancreas and liver (Figure 1). From these, the two protein-encoding genes Fms Related Receptor Tyrosine Kinase-1 (FLT1) and Ephrin type-B receptor-2 (EPHB2) overlapped in two or more pancreatitis phenotypes in the TWAS and GWAS, respectively. EPHB2 was associated with a 8.6-fold (P=1.84 x 10-8) and a 31.4-fold (P=2.87x 10-8) higher likelihood of pancreatic injury and TIAP, respectively.

Figure 1.

Conclusion

FLT1—a gene that encodes a receptor tyrosine kinase and is a member of the vascular endothelial growth factor receptor (VEGFR) family—and EPHB2—a gene that encodes a member of the Eph receptor family, which is the largest subgroup of the receptor tyrosine kinase family—are novel genetic markers associated with pancreatitis in patients taking AZA. VEGF can potentiate inflammation and the pancreas microenvironment is known to promote VEGF expression, which has been linked to pancreatic cancer development; anti-VEGF treatments have been investigated both for mitigating inflammation and also anti-pancreatic cancer treatment. Future studies validating our findings in AZA-induced pancreatitis are warranted.

References

[1]Crockett et al. Gastroenterology (2018). 154(4):1096-1101.

[2]Heap et al. Nature Genetics (2014). 46:1131-1134

Disclosure of Interests

Shailja Shah Consultant of: ad hoc consultant for Phathom pharmaceuticals, Tyler Reese: None declared, Jacy Zanussi: None declared, Alyson Dickson: None declared, Laura Daniel: None declared, Ran Tao: None declared, Tyne Miller-Fleming: None declared, Peter Straub: None declared, Adriana Hung: None declared, Puran Nepal: None declared, Wei-Qi Wei: None declared, Elizabeth Phillips: None declared, Nancy Cox: None declared, Charles M. Stein: None declared, QiPeng Feng: None declared, Cecilia P. Chung: None declared

Ninety-day and one-year healthcare utilization and costs after knee arthroplasty

OBJECTIVES

This study examined ninety-day and one-year postoperative healthcare utilization and costs following total knee arthroplasty (TKA) from the health sector and patient perspectives.

DESIGN

This study relied on: 1) patient-reported medical resource utilization data from diaries in the Knee Arthroplasty Pain Coping Skills Training (KASTPain) trial; and 2) Medicare fee schedules. Medicare payments, patient cost-sharing, and patient time costs were estimated. Generalized linear mixed models were used to identify baseline predictors of costs.

RESULTS

In the first ninety days following TKA, patients had an average of 29.7 outpatient visits and 6% were hospitalized. Mean total costs during this period summed to $3,720, the majority attributed to outpatient visit costs (84%). Over the year following TKA, patients had an average of 48.9 outpatient visits, including 33.2 for physical therapy. About a quarter (24%) of patients were hospitalized. Medical costs were incurred at a decreasing rate, from $2,428 in the first six weeks to $648 in the last six weeks. Mean total medical costs across all patients over the year were $8,930, including $5,328 in outpatient costs. Total costs were positively associated with baseline Charlson comorbidity score (P < 0.01). Outpatient costs were positively associated with baseline Charlson comorbidity score (P = 0.03) and a bodily pain burden summary score (P < 0.01). Mean patient cost-sharing summed to $1,342 and time costs summed to $1,346.

CONCLUSIONS

Costs in the ninety days and year after TKA can be substantial for both healthcare payers and patients. These costs should be considered as payers continue to explore alternative payment models.

Differences in health outcomes associated with initial adherence to oral antidiabetes medications among veterans with uncomplicated Type 2 diabetes: a 5-year survival analysis

AIM

To determine the association of adherence to oral antidiabetes medication with macrovascular and microvascular complications, time to insulin therapy, revascularization, admissions, and death among veterans with uncomplicated diabetes.

METHODS

This was a retrospective cohort study using the Veterans Affairs Corporate Data Warehouse to examine 159 032 veterans diagnosed with uncomplicated diabetes during 2002-2014 and starting oral antidiabetes therapy for the first time. The first uncomplicated diabetes diagnosis was identified and confirmed by subsequent oral antidiabetes therapy initiation. Adherence was calculated from outpatient pharmacy records using the proportion of days covered over the first year of therapy. Health outcomes were observed up to 5 years beyond the first oral antidiabetes agent fill, and compared according to adherence status using Cox proportional hazards models adjusted for baseline demographic and clinical characteristics.

RESULTS

During the first 5 years of oral antidiabetes treatment, people initially non-adherent to oral antidiabetes therapy were more likely to experience myocardial infarction (hazard ratio 1.14, 95% CI 1.03-1.27) and ischaemic stroke (hazard ratio 1.22, 95% CI 1.05-0.1.42), or to die (hazard ratio 1.21; 95% CI 1.15-1.28). Veterans with <20% adherence to oral antidiabetes therapy in the first year had particularly high hazards for ischaemic stroke (hazard ratio 1.78, 95% CI 1.27-2.49) and all-cause death (hazard ratio 1.33, 95% CI 1.17-151). Adherent people were more likely to be diagnosed with a microvascular complication or chronic kidney disease.

CONCLUSIONS

People who are non-adherent to treatment were more likely to experience detrimental health outcomes within the first 5 years of antidiabetes therapy. Adherence is paramount to disease management and this should be stressed from the time at which treatment is initiated.

Pre-treatment PSA velocity (PSAV) is not predictive of outcomes after radical prostatectomy (RP) in a referral cohort

4619

Background: PSA velocity has recently been reported as a significant predictor of prostate cancer-specific mortality in early-staged cancer patients selected from a large screening trial where the median PSA was 4.3ng/ml. In order to determine the broad applicability of these data to a more general population, we examined the effect of PSAV in a referral cohort. Methods: We analyzed over 643 prostate cancer cases diagnosed from 1995–2005. 196 patients underwent RP as definitive therapy. One patient received neoadjuvant hormonal therapy and was excluded. Our goal was to identify whether PSAV prior to RP was predictive of Relapse-Free Survival (RFS) and Overall Survival (OS). 156 patients had at least 2 pre-op PSA values available for the calculation of PSAV (median = 3 values/patient). PSAV was calculated by linear regression analysis using all pre-treatment PSA values available. Results: With a median f/u of 54 months, there were 14 deaths and 26 patients had biochemical relapse. At 5 years, OS was 93% and RFS was 91%. Surprisingly, no patients with biochemical relapse have died, and thus prostate cancer-specific mortality is 0%. In addition to the variables in the table below, seminal vesicle involvement (p = 0.015) and margin positivity (p=0.005) were also significantly associated with RFS confirming the expected prognostic factors in our cohort. On multivariate analysis, Gleason Grade was the only significant predictor of overall survival. For RFS, Gleason Grade, margin involvement, and pre-treatment PSAs were the only significant variables. Neither PSAV nor PSA Doubling Time were significantly associated with RFS or OS by univariate or multivariate analysis. Conclusions: Pre-treatment PSAV is not significantly associated with prostate cancer outcomes in a heterogenous population referred for an elevated PSA. Gleason Grade is still the most significant predictor of overall survival.

[Table: see text]

No significant financial relationships to disclose.

Temporal and spatial expression of murine acid beta-glucosidase mRNA

The hydrolysis of glucosylceramide (GC) to ceramide and glucose requires the action of the lysosomal enzyme, acid beta-glucosidase (GCase), encoded by gba in the mouse. Gaucher disease, an autosomal recessive disorder, results from the inherited deficiency of this enzyme. Although enzyme activity is present in all mammalian tissues, the patterns of mRNA expression have not been explored. In situ hybridization analyses of mouse embryonic, newborn, and adult tissues were conducted to evaluate the spectrum of gba mRNA expression. Signals were present in all tissues and cell types. Distinct patterns of differential expression were identified in specific tissues and cell types, and at defined developmental stages. Differential expression was first observed around E14 in the intestinal tract, kidneys, skeletal system, and skin. At E18, moderate intensity signals were in adipocytes of brown fat and pancreatic cells. Differential expression remained in skin, bone, and the GI tract postnatally. In the postnatal and adult animals increasing expression was observed throughout the CNS, esophageal epithelium, intestinal villi, pancreas, and thymus and lymph node capsular cells. These tissue-, cell-, and developmental stage-specific variations of the gba mRNA level indicate major developmentally regulated changes in the expression pattern of gba in the late gestational period and postnatally.

A prospective randomized study to determine the optimal dose of intravenous vitamin K in reversal of over-warfarinization

The optimal management of asymptomatic overanticoagulated patients remains unknown. We measured international normalized ratio (INR), activated partial thromboplastin time (APTT) and prothrombin fragment 1 + 2 (F1 + 2) over 7 d in 24 asymptomatic or mildly haemorrhagic patients on warfarin with prolonged INR of > 7.0 who were randomized to receive 0.5 mg, 1 mg or 2 mg intravenous vitamin K. Of six severely overanticoagulated patients (INR > 9.5 with APTT ratio > 2), five failed to achieve an INR < or = 4.0 on day 1, irrespective of vitamin K dose given. In the remaining 18 cases, an optimal response (INR 2-4 at day 1) was observed in 67% of those receiving 0.5 mg vitamin K, but only in 33% of those receiving 1 or 2 mg, the majority of whom developed an INR < 2.0. Our results support an optimal dose of 0.5 mg i.v. vitamin K for most overanticoagulated patients, with possibly a repeat dose in the small group of severely overanticoagulated patients.

Cloning of rat and human inducible penile nitric oxide synthase. Application for gene therapy of erectile dysfunction

Erectile dysfunction is mainly due to the inability of the cavernosal smooth muscle of the penis to undergo complete relaxation. In the aging rat model, erectile dysfunction is accompanied by a reduction of penile smooth muscle compliance and, in very old animals, by a decrease in penile nitric oxide synthase (NOS), which is responsible for the synthesis of the mediator of penile erection, nitric oxide (NO). We have investigated whether the stimulation of penile NOS expression by local induction or gene therapy can mitigate erectile dysfunction in the aged rat. A mix of iNOS (inducible NOS) inducers was continuously delivered to the penises of 5- ("adult"), 20- ("old"), and 30- ("very old") mo-old rats for 3-6 days, and the erectile response to electrical field stimulation of the cavernosal nerve was measured. The erectile dysfunction observed in old and very old rats as compared to adult animals was ameliorated by treatment with iNOS inducers. Penile iNOS was detectable in the penis of these rats by Western blot, NADPH diaphorase, and NOS activity assays. Inducible NOS was inducible in vitro in both rat and human corpora cavernosal tissue and in rat penile smooth muscle cells (RPSMC), as shown by Western blots. However, NO synthesis in cavernosal tissue upon iNOS protein induction remained low, indicating that the increased NOS levels were under physiological control. The iNOS cDNA was cloned from induced RPSMC mRNA and generated by reverse transcriptase polymerase chain reaction (RT-PCR) from induced human penile smooth muscle cells and corporal tissue. The coding regions from both the rat (RPiNOS) and human (HPiNOS) penile iNOS showed several amino acid differences from their analogous isoform in nonpenile tissues. RPiNOS cDNA injected into the penis mitigated the aging-associated erectile dysfunction. The iNOS construct was detected in cavernosal tissue by PCR, and its expression by RT-PCR and Western blots. These results open the way for the possible use of NOS isoforms in the management of erectile dysfunction.

Expression of inducible nitric oxide synthase in smooth muscle cells from rat penile corpora cavernosa

Nitric oxide (NO), the main mediator of penile erection, is assumed to be synthesized in the penis by the neuronal constitutive nitric oxide synthase (nNOS). However, nNOS has not been identified in the penile smooth muscle, the target of NO action. The other NOS isozymes, the inducible NOS (iNOS) and the endothelial NOS (eNOS) have not been reported in any penile tissue. The smooth muscle vascular and trabecular tissue from rat corpora cavernosa is represented in vitro by cell cultures designated RPSMC. To determine whether iNOS can be expressed in penile smooth muscle, RPSMC were treated with different lymphokines and/or bacterial lipopolysaccharide (LPS). The selected inducer, LPS/interferon, elicited at 48 hours up to a 50-fold increase in nitrites in the medium; the nitroarginine methyl ester (L-NAME), aminoguanidine, actinomycin D, cycloheximide, transforming growth factor-beta1 (TGF-beta1), and dexamethasone, but was resistant to nifedipine and platelet-derived growth factor AB (PDGF-AB). iNOS induction increased with cell passage. The [3H]L-arginine/citrulline measurement of NO synthesis with intact cells confirmed these results. Incubations of soluble and particulate fractions showed that the cytosol contained most of the activity (Km = 43 microM), which was partially inhibited by ethyleneglycal-bis-tetraacetic acid (EGTA). The 4.4-kb iNOS mRNA peaked at a late period (24-30 hours) and remained high for up to 72 hours. iNOS mRNA induction was strongly inhibited by actinomycin D and dexamethasone, partially inhibited by TGF-beta1, inhibited slightly by PDGF-AB, and unaffected by nifedipine. These results show that iNOS can be expressed in RPSMC in a cell passage-dependent fashion that has so far not been reported for other cell lines, and that the induction reaches much higher levels than in rat or human vascular smooth muscle cells. The expression pattern is also distinctive for the penile cells in time course of induction, Ca2+ dependence, response to certain agents, and mRNA stability.

Isolation of two pyruvate kinase activities in the parasitic protozoan Leishmania mexicana amazonensis

Using phosphocellulose affinity chromatography we were able to separate two pyruvate kinase (EC 2.7.1.40) activities in the parasitic protozoan Leishmania mexicana amazonesis. One activity (PYK1) showed hyperbolic kinetics and was decreased by fructose 2,6-bisphosphate, whereas the second activity (PYK2) showed sigmoidal kinetics for the substrate phosphoenolpyruvate and was activated by fructose 2,6-bisphosphate. Molecular sieve chromatography (Sephacryl S-400) of PYK1 produced a single peak of apparent molecular mass around 200,000, while PYK2 eluted at a position corresponding to M(r) 55,000.

In vivo selected promoter and ribosome binding site up-mutations: demonstration that the Escherichia coli bla promoter and a Shine-Dalgarno region with low complementarity to the 16 S ribosomal RNA function in Bacillus subtilis

We have constructed a plasmid, pQS1, in which a mouse dihydrofolate reductase (5,6,7,8-tetrahydrofolate:NADP:oxidoreductase; EC 1.5.1.3; DHFR) cDNA is inserted in the unique PstI site of a gram-positive/gram-negative shuttle vector derived from pBR322. The cDNA is expressed under the control of the bla promoter, which, like most gram-negative bacterial genes, is considered not to be expressed in Bacillus subtilis, and its coding sequence is translated from a polycistronic message. We have selected in vivo and studied, in Escherichia coli and B. subtilis, expression mutants with promoter and ribosome binding site sequence mutations. One promoter mutation changes the third nucleotide of the -35 region from a C to a G. As expected, this substitution results in increased transcriptional activity in E. coli. In B. subtilis, this mutation induces the accumulation not only of a low but significant amount of dhfr mRNA but also of DHFR, demonstrating that binding strengths with a free energy as low as -9.4 kcal/mol are sufficient to promote ribosome binding in B. subtilis. The association of the promoter mutation (C-G) with a mutation which creates a strong B. subtilis ribosome binding site (-21 kcal/mol) results in the accumulation of a large amount of dhfr mRNA. This demonstrates the importance of having an efficient ribosome binding site in the evaluation of promoter function: for example, with this strong ribosome binding site we can show that the wild-type bla promoter is recognized by the B. subtilis transcription machinery.

Similarities between the regulatory sequences of the unrelated tetracycline genes of pBR322 and Tn10

The regulatory regions of the tetracycline genes present in pBR322 (pSC101) and in the transposon Tn10 are compared. They show a low degree of nucleotide sequence similarity but a high level of structure similarity. Furthermore, analyses of RNAs transcribed in the opposite direction of the pBR322 tet gene show that there are two mRNA initiation sites separated by 29 nucleotides. This suggests the existence of two promoters for the tet repressor gene in Tn10. These features reveal a strong resemblance of the mode of regulation between the tet operons of Tn10 and pSC101.

Expression of the mouse dihydrofolate reductase cDNA in B. subtilis: a system to select mutant cDNAs coding for methotrexate resistant enzymes

With the aim to obtain a cDNA coding for a mammalian methotrexate resistant dihydrofolate reductase (Dhfr) a plasmid ( pQS1 ) harboring the mouse wild type Dhfr cDNA was constructed and used to transform a methotrexate sensitive bacteria: B. subtilis. A plasmid, pQS4 , expressing large amount of Dhfr in both E. coli and B. subtilis was isolated through a two steps selection with two substrate analogues, trimethoprim followed by methotrexate. This new plasmid has a 54 bp duplication including the beta-lactamase promoter and a deletion of 564 bp removing the 5' end of the beta-lactamase coding region. These changes create a new -35 region TTGAAA and a potentially stronger binding site for both E. coli and B. subtilis 16S ribosomal RNA. pQS4 transformed B. subtilis were then grown in the presence of high level of methotrexate and resistant mutants isolated. One of them, pQS6 , which codes for an enzyme about 50 times more resistant to methotrexate than the wild type Dhfr was sequenced. It shows that a point mutation replaces the glutamine residue at position 35 by a proline.