DISORDER

CROSS-REFERENCE TO RELATED APPLICATION This application claims priority to US Provisional Application No. 61/757,855, filed on January 29, 2013, the content of which is incorporated herein by its entirety.

BACKGROUND

Bipolar disorder is a disabling mental illness characterized by episodes of both elevated or irritable mood and depression. See, e.g., Muller-Oerlinghausen et al., Lancet 2002; 359:241-7; and Frye MA., N Engl J Med 2011; 364:51-9. Currently, lithium is the first-line choice for maintenance treatment in bipolar disorder and reduces risk of relapse and suicide. See, e.g., Fountoulakis et al., Eur Arch Psychiatry Clin Neurosci 2012; 262 Suppl 1 : 1 -48. However, many patients do not respond to lithium treatment.

SUMMARY

Described herein is a method of determining lithium responsiveness in a bipolar disorder patient. The method includes: obtaining a sample from a patient having bipolar disorder; and assaying the sample for the presence or absence of one or more glutamate decarboxylase-like 1 (GADL1) gene variants selected from the group consisting of: (i) a T allele of the single nucleotide polymorphism (SNP) rsl7026688; (ii) a G allele of the SNP rs 17026651 ; (iii) GADL1 IVS8+48delG; (iv) a genetic variant that is associated or in linkage disequilibrium with SNP rsl7026688, SNP rsl7026651, or GADL1 IVS8+48delG, or a combination thereof; and (v) a haplotype containing or in linkage disequilibrium with SNP rsl7026688, SNP rsl7026651, or GADL1 IVS8+48delG, or a combination thereof. The presence of one or more of the GADL1 gene variants indicates that the patient is responsive to lithium treatment.

The sample can be a genomic DNA sample, RNA sample, cDNA sample, or protein sample obtained from a tissue or bodily fluid of the patient, e.g., blood or saliva. The assaying step can be performed by, for example, DNA sequencing, restriction enzyme digest, polymerase chain reaction (PCR), hybridization, real-time PCR, reverse transcriptase PCR, or ligase chain reaction, and an immunoassay. In one embodiment, the patient is a bipolar I disorder patient of Han Chinese descent. Also described is a method of treating bipolar disorder in a patient. The method includes determining whether a patient carries one or more of the above-described GADL1 gene variants. If the patient carries one or more of the gene variants, the patient can be administered lithium treatment.

Further, described below is a kit for determining whether a subject carries one or more of the GADL1 gene variants described herein. The kit includes one or more probes for detecting the GADL1 variants.

The details of one or more embodiments are set forth in the description below. Other features, objects, and advantages of the embodiments will be apparent from the description and from the claims.

DETAILED DESCRIPTION

It was unexpected discover that certain variants of glutamate decarboxylase-like 1 (GADL1) are associated with lithium responsiveness in bipolar patients. A patient carrying one or more of these GADL1 variants is more likely to be responsive to lithium treatment.

GADL1 belongs to the group II decarboxylase family. The genomic location

(chromosome 3), genomic DNA sequences (see, e.g., NC_000003.11 Reference GRCh37.pl3 Primary Assembly, chr3:30767692-30936153), cDNA sequences (see, Accession No.

NM_207359.2), and protein sequences of human GADL1 (see, e.g., Accession No.

NP_997242.2) are known in the art.

The amino acid sequences of exemplary GADL1 polypeptides and cDNA sequences encoding the polypeptides are shown below.

Two single nucleotide polymorphism (SNPs), rsl7026688 and rsl7026651, located in the introns of the GADL1 gene, and a 1-base deletion variant, IVS8+48delG, were found to be associated with lithium responsiveness. IVS8+48delG is a G deletion located in intron 8 of the GADL1 gene. Bipolar patients carrying the T allele of rs 17026688, the G allele of rs 17026651, and/or IVS8+48delG are more likely to respond to lithium treatment.

Described herein is a method of determining or predicting whether a bipolar patient is a good responder of lithium treatment. A bipolar patient (e.g., a bipolar I patient) can be identified using methods known in the art and described herein.

For example, the presence of the T allele of rs 17026688 (e.g., genotype TT or CT) in a bipolar patient indicates that the patient is a good responder of lithium treatment. On the other hand, a patient carrying the C allele of rsl7026688 (i.e., genotype CC) is less likely to have a good response to lithium. In another example, a bipolar patient carrying the G allele of rsl7026651 (e.g., genotype GG or CG) is more likely to be a good responder of lithium treatment than a patient carrying the C allele (e.g., genotype CC). In yet another example, the presence of the IVS8+48delG variant in a bipolar patient indicates that the patient is a good lithium responder.

If a bipolar patient has one or more genetic variants associated with good lithium response, the patient can be administered lithium treatment. For a bipolar patient who lacks such variants, alternative therapeutics may be preferred. Alternative bipolar therapeutics include, but are not limited to, valproate, carbamazepine, oxcarbazepine, and lamotrigine.

Other GADL1 genetic markers may also be used to determine a bipolar patient's responsiveness to lithium. For examples, such genetic markers include known SNPs located in the GADL1 gene and those GADL1 variants described herein.

The presence of a genetic variant can be determined by direct detection of that variant or a particular region within it. Genomic DNAs for allele detection can be prepared from a patient by methods well known in the art. Detection of a region within a genetic marker of interest includes examining the nucleotide(s) located at either the sense or the anti-sense strand within that region. Methods known in the art can be used to detect a particular region or genetic variant, e.g., sequencing, sequence specific oligonucleotide-hybridization , realtime PCR, ligase chain reaction, or CSSO-ELISA (M. Hiratsuka et al, J. of Biochemical and Biophysic. Methods, 67:87-94, 2006).

The presence of an allele of interest also can be determined by detecting genetic markers equivalent to the allele. Genetic markers near the allele of interest tend to co- segregate, or show a linkage disequilibrium, with the allele. Consequently, the presence of these markers (equivalent genetic markers) is indicative of the presence of the allele of interest, which, in turn, is indicative of lithium responsiveness.

Alternatively or in addition, RNAs, cDNAs, or protein products of alleles of interest can be detected to determine the presence or absence of the alleles. For example, the IVS8+48delG variant results in a truncated form of GADL1 lacking exons 7 and 8. Mass spectrometry assays, e.g., MALDI-MS, LC-MS, and LC-MS/MS, and immunoassays, e.g., ELISA, Western blot, radioimmunoassay (RIA), fluorescent immunoassay (FIA), and luminescence immunoassay (LIA), can be used to determine the protein product of an allele. Reverse transcriptase PCR cab used to detect the mRNA product of an allele.

Genomic DNA, cDNA, RNA, or protein samples from patients can be prepared from various tissues and bodily fluids of the patients, e.g., blood, saliva, urine, and hair.

Also described herein is a kit containing probes for detecting one or more genetic markers, e.g., rsl7026688, rsl7026651, and IVS8+48delG. The term "probe" used herein refers to any substance useful for detecting another substance. Thus, a probe can be an oligonucleotide or conjugated oligonucleotide that specifically hybridizes to a particular region within an allele of interest. The conjugated oligonucleotide refers to an

oligonucleotide covalently bound to chromophore or a molecules containing a ligand (e.g., an antigen), which is highly specific to a receptor molecular (e.g., an antibody specific to the antigen). The probe can also be a PCR primer, together with another primer, for amplifying a particular region within the allele of interest. Further, the probe can be an antibody that recognizes an allele of interest or a protein product of the allele. Optionally, the kit can contain a probe that targets an internal control allele, which can be any allele presented in the general population, e.g. GAPDH, β-actin, KIR. Detection of an internal control allele is designed to assure the performance of the kit. The probes can be immobilized on a solid support, e.g., an array. The kit can further include tools and/or reagents for collecting biological samples from patients, as well as those for preparing genomic DNA, cDNAs, RNAs or proteins from the samples. For example, PCR primers for amplifying the relevant regions of the genomic DNA may be included.

In one example, the kit contains a first probe, a second probe, and a third probe, each for detecting the T allele of rsl7026688, the G allele of rsl7026651, or IVS8+48delG. These probes can each be a pair of PCR primers or a labeled oligonucleotide useful in hybridization assays. Optionally, the kit can include an additional probe for detecting an internal control allele.

The specific example below is to be construed as merely illustrative, and not limitative of the remainder of the disclosure in any way whatsoever. Without further elaboration, it is believed that one skilled in the art can, based on the description herein, utilize the present disclosure to its fullest extent. All publications cited herein are incorporated herein by reference in their entirety.

We selected subsets of participants from a sample of 1761 persons of Han Chinese descent with bipolar I (BPI) disorder recruited by the Taiwan Bipolar Consortium. We assessed response to lithium treatment using the Alda Scale, and carried out a genome-wide association study (GWAS) on 294 BPI patients receiving lithium treatment. The SNPs showing strongest association with response to lithium were then tested for association in a replication sample of 100 patients and further tested in a follow-up series of 24 patients. We sequenced the exons, exon-intron boundaries and part of the promoter of GADLl in 94 responders and 94 non-responders from the GWAS sample.

We found that two SNPs in high linkage disequilibrium, rs 17026688 and rs 17026651, located in the introns of glutamate decarboxylase-like 1 (GADLl), showed the strongest associations in the GWAS (P = 5.50 X10 ^-37 and 2.52 X 10 ^-37 ) and the replication set of 100 patients (P=9.19 X10 ^-15 for each SNP). These two SNPs had a sensitivity of 0.93 for predicting lithium response and differentiated between the good and poor responders in the follow-up cohort. Re-sequencing of GADLl disclosed a novel variant, IVS8+48delG, which lies in intron 8 of the gene and is in complete linkage disequilibrium with rs 17026688.

IVS8+48delG was shown to affect splicing.

Our study showed that rsl7026651, GADLl IVS8+48delG, and rsl7026688 are useful biomarkers in predicting response to lithium therapy in persons with BPI disorder. These alleles are rare in persons of European and African ancestry. Other variants in GADL1 may influence response to lithium therapy in these populations.

METHODS

PARTICIPANTS

The study was conducted by Taiwan Bipolar Consortium established in 2003 with members from the Institute of Biomedical Sciences, Academia Sinica and 25 psychiatric departments of general hospitals and psychiatric institutions in Taiwan. The Consortium initially set out to understand genetic susceptibility to BPI disorder, and broadened its scope to the pharmacogenetic study of mood stabilizers. The first part of the study has been described previously. See, Lee et al., Mol Psychiatry 2011 ; 16:548-56. In brief, unrelated patients with BPI, aged 20 to 65, were recruited from the psychiatric departments and institutions of the Taiwan Bipolar Consortium. All of these patients had been diagnosed according to DSM-IV criteria for BPI disorder with recurrent episodes of mania with or without depressive episode(s). We excluded patients with other psychotic and affective disorders.

Psychiatric nurses and psychiatrists evaluated study participants using a cross- culturally validated Chinese version of the Schedules for Clinical Assessment in

Neuropsychiatry (SCAN) (see, Cheng et al., Br J Psychiatry 2001; 178:567-72),

supplemented by available medical records and reports from family members and in-charge psychiatrists. Only patients of Han-Chinese descent were considered for the study (ancestry was determined by verbal report of patients to members of the research team). We recruited 1761 BPI patients between March 2003 and the end of May 2012.

STUDY DESIGN AND OVERSIGHT

We carried out a "discovery" GWAS, and two tests of replication. For GWAS, we first identified 294 from 1647 (17.9%) persons with BPI consecutively recruited from outpatient clinics and inpatient units of the 25 psychiatric departments and institutions in the Taiwan Bipolar Consortium. We genotyped each of the 1647 patients by array (see below). The 294-patient GWAS sample had received lithium prophylaxis treatment with good adherence for at least two years. The remainder of the patients in the series did not. We identified genetic regions associated with response to lithium prophylaxis treatment, and then carried out a test of replication using SNPs marking these loci in an independent group of 100 persons with BPI disorder. These 100 patients were selected from 114 BPI patients (distinct from those in the sample of 1647 patients) referred to us by staff psychiatrists who had treated these patients for more than 10 years with lithium and observed good drug compliance. Fourteen of the 114 patients were excluded because they did not fulfil our inclusion criteria (see Phenotype Definition and Assessment below).

In a second test of replication, we genotyped an independent series of 24 patients who had received lithium monotherapy for at least 2 years, through May 2012. We based their inclusion on a life chart (see Phenotype Definition and Assessment below) constructed for all patients in the 1647 -patient sample. Each of these 24 patients had good drug adherence to mood stabilizer(s) other than lithium prior to their commencement of lithium monotherapy but with unsatisfactory response.

The study was approved by the institutional review boards of participating hospitals and Academia Sinica, Taiwan, and we obtained written informed consent from all participants.

PHENOTYPE DEFINITION AND ASSESSMENT

To assess response to long-term lithium prophylaxis treatment in BPI disorder, we prepared a life chart with graphic depiction of lifetime clinical course for each of the BPI patients recruited before June 2012 (N=1761). This life chart included all manic, hypomanic, and depressive episodes with date of onset (year and month), duration, and severity

(including the extent of functional disability, hospitalization, and the presence of psychotic features), all doses of and duration of treatment with psychotropics and mood stabilizers known have been prescribed, drug adherence recorded in medical chart for all visits at outpatient clinics, all recorded blood levels of mood stabilizers, and any adverse drug reactions. We depicted this information graphically, based on integrated information gathered from direct interview with patients and their family members, interviews with in-charge psychiatrists, and a thorough medical chart review. Based on this life chart, we determined whether individual patients had good drug adherence.

The phenotype of lithium response was assessed based on the life chart, using the Retrospective Criteria of Long-Term Treatment Response in Research Subjects with Bipolar Disorder developed by Martin Alda and Colleagues (Alda Scale). The Alda scale has two criteria. Criterion A measures the extent of clinical improvement in illness activity and takes into account the frequency, duration, and severity of episodes during periods of lithium treatment considered adequate in duration and dosage, compared with the frequency, duration, and severity of episodes during periods off the lithium treatment, on a scale from 0 (no change, or exacerbation of disease severity) to 10 (complete remission). Criterion B (B1-B5, each rated as 0, 1 or 2 points) is used to establish whether there is a causal relationship between clinical improvement and the treatment. B 1 and B2 indicate the recurrence risk off the lithium treatment [number (Bl) and frequency (B2) of episodes]. B3 is a measure of the length of lithium treatment. B4 indicates compliance with lithium. B5 is a measure of concomitant psychotropic medication during periods of stability. The total score is obtained by subtracting the sum of the B scores from the A score.

In investigating a causal relationship between lithium treatment and clinical improvement of individual patients, it is important to ensure (i) comparability in clinical course between on-lithium and off-lithium periods (number, frequency and severity of episodes); (ii) satisfactory drug compliance, and (iii) the minimization of influence from additional medications (hypnotics, antidepressants, antipsychotics, and other mood stabilizers). We determined inclusion criteria accordingly, to minimize the misclassification of responders and non-responders. For example, we included patients showing poor response to lithium combined with prolonged use of antipsychotics or additional mood stabilizers, and excluded patients showing good response to lithium combined either with additional mood stabilizer throughout the course, or with prolonged use of high-dosage antipsychotics (B5=2). As shown in Table 1, nearly all of the study subjects (95.0%) with a B5 score of 2 were poor responders to treatment at the optimal cutoff of 5/6.

For patients in the second test of replication, we carried out regular follow-up evaluations (usually monthly and at least once every 3 months), including assay of lithium to assess drug adherence, and a SCAN interview to assess clinical condition, at outpatient clinics for at least 2 years.

An inter-rater reliability of the Alda Scale was carried out with 18 randomly selected BPI patients from the GWAS group. Three senior psychiatrists performed ratings based on the life chart. We observed an intraclass correlation among the 3 raters of 0.904 (see Table 2) for the total score (0-10). OUTCOMES

Previous studies using the Alda Scale have adopted a total score of 6/7 as the optimal cut-off point between non-responders (0-6) and responders (7-10) in lithium prophylaxis treatment. See, e.g., Grof et al., J Clin Psychiatry 2002; 63:942-7; and Squassina et al, Pharmacogenomics 2011 ; 12:1559-69. For the GWAS, we selected 4 potential cutoff points at 4/5, 5/6, 6/7 and 7/8 for classifying subjects with, respectively, greater than 50%, 65%, 80% and 90% reduction of illness activity as responders in lithium prophylaxis treatment.

GENOTYPING, IMPUTATION, AND SEQUENCING

We genotyped the 1647 participants using Illumina HumanHap550-Duo BeadChip and HumanOmnil-Quad BeadChip and integrated the two data sets through imputation with HapMap Phase 2 data. Quality control procedures were applied to the genotype data and the imputed data. We genotyped the top SNPs in the two replication series using SEQUENOM MassARRAY, and then sequenced GADLl in 94 responders and 94 non-responders randomly selected from the GWAS group using Applied Biosystems 3730 DNA Sequencer.

STATISTICAL ANALYSIS

We compared the prevalence of alleles, implicated by GWAS, in non-responders and responders using the Cochran- Armitage trend test. The threshold P-value was set at 6.9 X 10- ⁹ after a Bonferroni corrected for the number of SNPs (1,814,186) and for the 4 different cut points. We examined P-value distributions using quantile-quantile (Q-Q) plots. We analyzed the GWAS data according to the 4 cutoff points to classify non-responders and responders. We also evaluated the top hits with adjustments for psychotic features (delusion and hallucination), family history of BPI in first-degree relatives, rapid cycling, age at onset, sex, and history of alcoholism using PLINK v. 1.07.

RESULTS

THE STUDY PARTICIPANTS

Demographic and clinical characteristics of the 394 study participants of the GWAS and Replication set are shown in Table 3, which includes clinical phenotypes previously reported to be associated with lithium responses. The median age was 49 years old and males and females were similar in proportion. A high proportion of study subjects were found to have a history of psychotic features (60%). The percentages of rapid cycling and family history of BPI in the first-degree relatives were 25% and 31%, respectively. Early-onset disease occurred in 15% and a history of alcoholism in 8% of the study participants.

During off-lithium periods, the median number of episodes was 6 with a range of 4 to 144 and the median frequency of episodes was 1 per year with 0.4 (1 per 2.5 years) as the lowest. The median duration of lithium prophylaxis therapy with good adherence among study patients was 7 years, the shortest being 2 years. The reported lithium blood levels were equal to or exceeded 0.5 mM. We compared disease activity during periods with good adherence to activity during off-lithium periods. We observed that a high proportion of participants (88%) were found to be taking other drugs, and a quarter of them had received a prolonged course of antidepressant, antipsychotic, and/or mood stabilizer, in addition to lithium. See Table 4 for the distribution of the Alda Scale scores in the GWAS and replication groups.

ASSOCIATION ANALYSIS

We did not observe substantial population stratification nor cryptic genetic

relationships among the 294 subjects analyzed by GWAS. SNPs on chromosome 3p24.1 showed association with response to lithium at genome-wide significance (P < 6.9 X 10 ^-9 ); no other chromosome region showed an association with genome-wide significance (Summary statistics from the genomewide association study can be found in the database of Genotypes and Phenotypes [dbGaP] of the National Center for Biotechnology Information, accession number phs000692.vl.pl.). Two SNPs in particular, rsl7026688 and rsl7026651, located in the introns of GADLl, encoding glutamate decarboxylase-like 1, approximately 7.2 kilobases apart, showed the strongest associations, with a cutoff point of 5/6 (P=5.50 X 10 ^~37 and 2.52 X 10 ^~37 , respectively). The highest sensitivity and specificity for response to lithium were 0.93 and 0.85 for rsl7026688, and 0.93 and 0.86 for rsl7026651 at this cutoff point (Table 5).

We then genotyped rs 17026688 and rs 17026651, together with flanking SNPs that showed genomewide significance in the GWAS, in GADL1 in the replication group. Both rs 17026688 and rs 17026651 showed the strongest associations in the test of replication (P=9.19 X 10 ^-15 for both SNPs). The distributions of allele prevalence were not significantly different for the top SNPs in the GWAS and replication samples. The P values of rs 17026688 and rs 17026651 in the combined series, comprising 394 study participants, were P =1.66 X 10 ^-49 and 7.07 X 10 ^-50 , respectively (Table 6).

Fisher's exact test for association between allele status and good responder status in the combined series yielded P values of 6.69 X 10 ^-62 and 8.30 x 10 ^-63 , respectively (Table 7). We observed the two SNPs to be in high or absolute linkage disequilibrium (In the GWAS, D' = 1.0 and r ² = 96.6%, and in the replication group, D' = 1.0 and r ² = 100%).

Among the 24 patients in the second test of replication (the follow-up study), the two top SNPs were also in complete linkage disequilibrium. All carriers of the "response" alleles (N=16) showed good response to lithium treatment (total Alda score ranged 8-10), and all of the non-carriers (N=8) showed poor response (total Alda score ranged 0-3) (Table 8). The GWAS (N=394) and the follow-up (N=24) study reached an acceptable power of 0.85 and 0.95, respectively.

We carried out further analyses of rs 17026688 (Table 9). To assess the influence of other factors that might contribute to lithium response, we performed logistic regression analyses on data from all 394 participants (Table 10). We found that the "response" allele T at rs 17026688 was associated with a much better response to lithium than was the alternative allele (P=3.39x10 ^-32 , OR=111.87, 95% CI 51.14-244.73, PPV=0.83, 95% CI 0.76, 0.88). In addition, patients with rapid cycling (irrespective of genetic status) had slightly better lithium response than those without (P= 8.47 x 10 ^-4 , OR=3.88, 95% CI 1.75-8.59). Under the same logistic regression model, the other 21 flanking SNPs showed nominal significance or non- significance when conditioned on rsl7026688 (all P values > 0.00185, Table 11). RE-SEQUENCING GADL1

Because the SNPs showing strongest association are in the introns of GADL1, we next looked for local variants likely to affect the expression of GADL1. Ninety-four responders and 94 non-responders were randomly selected from the 294-person GWAS set for sequence analysis of the exons, intron-exon boundaries, and a 2-kb region representing part of the promoter of GADL1. We found 32 genetic polymorphisms (Table 12), including a 1-base deletion in intron 8 of GADL1 (GAD LI IVS8 +48delG). We genotyped this variant in the all participants of each of the three parts of our study (N=417) and found it to be in complete linkage disequilibrium with rs 17026688.

EFFECT OF THE IVS8+48DELG VARIANT

To test the effect of IVS8+48delG variant on splicing of GADL1 mRNA, we determined the mRNA isoforms via RT-PCR in two glioma-derived neural cell lines. One line (GBM SlRl) carries a IVS8+48delG variant and the other (GBM8401) carries two non- mutant alleles. We detected two GADL1 mRNA isoforms: GBM8401 expressed the major mRNA isoform (499 bp) containing exons 5 to 10. We also detected the minor splice variant (364bp) containing exons 5, 6, 9 and 10 (exons 7 and 8 were omitted, presumably by alternative splicing). Compared with the cell line GBM8401, the cell line GBM SlRl (which carried the IVS8+48delG variant), showed low levels of the major isoform and elevated levels of the smaller alternatively-spliced mRNA species.

OTHER EMBODIMENTS

All of the features disclosed in this specification may be combined in any combination. Each feature disclosed in this specification may be replaced by an alternative feature serving the same, equivalent, or similar purpose. Thus, unless expressly stated otherwise, each feature disclosed is only an example of a generic series of equivalent or similar features.

From the above description, one skilled in the art can easily ascertain the essential characteristics of the described embodiments, and without departing from the spirit and scope thereof, can make various changes and modifications of the embodiments to adapt it to various usages and conditions. Thus, other embodiments are also within the claims.