Washington State, 2016¹³

United States

Study design: HTA

Literature search strategy: The authors searched PubMed (January 1, 2000 to August 15, 2016), OVID-Embase (1996 to 2016, week 33) and PsycINFO (1987 to July, week 4, 2016) databases for systematic reviews, meta-analyses, economic evaluations, primary clinical studies and practice guidelines; searches were updated to November 28, 2016

Included study characteristics:

14 studies, 12 guidelines included

Included studies published 2009–2016

Relevant studies: 4 clinical effectiveness studies

• Singh, 2015²³
• Winner, 2013²²
• Breitenstein, 2014²⁴
• Winner, 2015³⁵

3 economic evaluation studies:

• Perlis, 2009²⁶
• Olgiati, 2012²⁷
• Herbild, 2009²⁸

1 guideline

• WFSBP (Bauer, 2013)³²

Quality assessment tool: Hayes Evidence-Grading Guides and Quality Checklists

Objective: to determine the clinical utility, clinical effectiveness, and cost-effectiveness of genetic testing

N = NR

Included: Patient population was people of any age who were being prescribed medications for treatment of depression, mood disorder, psychosis, anxiety, ADHD, or substance use disorder. The interventions were clinical laboratory tests for genetic variants in targeted genes or in panels of genes. Test results were available to the medication prescriber in the experimental arm of the study. The settings were inpatient and outpatient settings.

Excluded: NR

Intervention:

GeneSight,^20–22 CNSDose,²³ ABCB1,²⁴ test for at least one of CYP2D6, CYP2C19, CYP2C9, and/or serotonin transporter genotype 5- HTTLPR,²⁵ HTR2A,²⁶ 5-HTTLPR,²⁷ CYP2D6,²⁸)

Comparator: usual care or no genetic testing

Outcomes: Patient Management: physician and patient decision-making regarding drug choice and/or dose; patient adherence to treatment regimen measured by percentage of drug therapy adherence; change in patient response to informed treatment (measured by treatment response with 50% reduction in HAMD-17, and remission with HAMD-17 <7, reduction in symptoms, Clinician Rated Quick Inventory of Depressive Symptomatology (QIDS-C16)) and change in adverse event rates as a result of informed treatment; cost-effectiveness measured by incremental QALY, or quality-adjusted life week (QALW) and incremental cost-effectiveness ratio (ICER), and recommendations regarding genetic testing

Length of follow-up: 5–12 weeks

Bousman, 2019⁶

Canada

Study design: SR with MA

Literature search strategy: The authors searched Embase, PsycINFO, Ovid Medline, ScienceDirect, Google Scholar and the Cochrane Central Register of Controlled Trials database up to May 2018. Studies published in English were included.

Included study characteristics:

5 RCTs included

Included studies published 2013–2018

Relevant studies: 5 RCTs

Quality assessment tool: Cochrane Risk of Bias Assessment Tool

Objective: To conduct a SR and MA of prospective, RCTs to examine the remission rates of pharmacogenetic-guided decision support tools relevant to depressive symptom remission in MDD

N = 1,737 patients

Included: adult participants (aged ≥18) diagnosed with MDD

Mean age: NR

Included interventions:

pharmacogenetic-guided treatment

Relevant Interventions:

unguided treatment

Outcomes:

Remission measured by HDRS-17 (score ≤ 7)

Length of follow-up: 8–12 weeks

Rosenblat, 2017¹⁴

Canada

Study design: SR with MA

Literature search strategy: The authors searched MEDLINE/PubMed and Google Scholar databases were searched from inception to October 2015 for published reviews, meta-analyses and primary studies.

Included study characteristics:

10 studies included

Included studies published 2009–2015

Relevant studies: 4 clinical effectiveness studies and 5 cost-effectiveness studies

Quality assessment tool:

Newcastle-Ottawa Scale

Objective: to determine the cost effectiveness and the effect on clinical outcomes of pharmacogenomic testing-guided treatment in the treatment of MDD as compared to unguided treatment

Included: adults (18–75 years old)

Mean age: NR

Excluded: NR

Included interventions:

pharmacogenomic testing-guided treatment

Relevant Interventions:

unguided treatment

Outcomes:

Depressive symptom severity, remission rate, cost- effectiveness

Length of follow-up: 8 weeks-3 months

Shan, 2019¹⁵

China

Study design: single-blinded RCT

Setting: inpatients and outpatients of Department of Psychiatry of the Second Xiangya Hospital, Central South University from September 2017 to July 2018.

Objective: to assess the effectiveness pharmacogenetic testing for medication therapy selection in patients with depression.

Inclusion criteria: Patients aged 18 to 51 years of age with a HAMD-17 total scores of > 17 at baseline and the first item of the HAMD-17 (depressive mood) > 2; never received psychiatric treatment or have interrupted antidepressant medication for more than 2 weeks (fluoxetine for at least 4 weeks); and with no psychotic symptoms

Excluded: having any other psychiatric diagnoses according to DSM-5; any physical illness such as liver and kidney diseases, cardiovascular diseases; any combination with other antipsychotic medications (both low and high doses), including typical and atypical antipsychotic and mood stabilizer; and pregnancy

Number of patients:

71 patients

n = 31 in guided group

n= 40 in unguided group

Mean age:

26.52 ± 7.92 years in guided group;

28.85 ± 8.93 years in unguided group

Intervention of interest:

pharmacogenetic testing for guided medication therapy selection

Comparator:

Unguided therapy selection

Relevant Outcome:

depression outcomes measured by HAMD-17 total score, response rate, remission rate, HAMA total score, blood routine, liver function, renal function, and electrocardiogram examination, adverse reactions measured by TESS

Length of follow-up: 8 weeks

Sluitter, 2019⁹

Netherl ands

Funding source: Grant by KNMP

Analysis: CUA

Approach: Model-based analysis; Markov model

Time horizon: 12 weeks

Cycle length: 12 weeks

Perspective: Netherlands societal

Discount rate: None

Analytic Approach: Univariate sensitivity analysis; scenario analysis

Adult patients with major depression (age ≥ 18 years) in primary care.

EM - one or two active alleles of CYP2D6 and a normal metaboliser activity;

PM - no active allele of CYP2D6, and a decreased metabolic activity

UM - more than two active alleles of CYP2D6 and increased enzymatic activity

Number of participants: Hypothetical cohort of 1000 patients was used in the model

Mean age: Not specified

Intervention: PGx

Comparat or: no PGx group = patients receiving standard care

-

Probabilities of being EM, PM or UM

-

Probabilities of SE for EM, PM or UM

-

Probabilities of having no SE for EM, PM, UM

-

After PGx, probability of PM having SEs

-

After PGx, probability of PMs having no SEs

-

Probability having a treatment effect on PM

-

Probabilities of having no treatment effect being PM, EM, or UM

-

Probabilities of having a treatment effect being PM, EM or UM

-

Probability of titration

-

Probability of switching to another class

-

Probability of waiting for treatment decision due to SE

-

Probability of suicide attempt

-

Probability of fatal suicide attempt

-

Associated with suicide attempts

-

Fatal suicide

-

PGx

-

GP visit

-

Psychiatrist visit

-

Due to labor loss

-

Prescription cost

-

Being depressed

-

No SEs

-

After titration

-

Switching

-

Waiting

-

Effect (remission)

-

No effect

-

Suicide attempt

-

Base case assumed no delay effect of therapy

-

Case of a failed attempt, a switch to a different drug was assumed, with outcomes similar to those of patients who did not show clinical improvement and who did not attempt suicide.

-

Concerning treatment effects, a general effect among depressive patients was assumed for the EM class.

-

Due to the lack of evidence for PM and UM metabolisers, assumptions on treatment effects had to be made.

-

For the PM group with an increased antidepressant blood level, the same treatment effect as for the EM group was assumed.

-

For the UM group with a decrease antidepressant blood level, a smaller treatment effect was assumed.

-

Probabilities of dosage titration, switching or waiting due to SE or no treatment effect were assumed to be equal for those without treatment effects, regardless of the metaboliser classes.

Groessl, 2018¹⁶

United States

Funding source: AltheaDx, manufact urer of the IDgenetix

Analysis: CEA

Approach: Model-based analysis; Markov model

Time horizon: 3 years

Perspective: US Societal

Discount rate: 2.5–3.5%

Analytic Approach: Univariate, one-way sensitivity analysis

Treatment naïve patients with MDD or patients with inadequately controlled MDD and a score of 20 or greater on the HAM-D17

Number of participants: 260 participants with moderate to severe depression from an RCT went through the Markov model.

Mean age: 48 years

Sex: 100% female

Intervention: IDGx

Comparat or: SOC

-

Clinical response measured by HAMD-17

-

Suicide rates for non-responders and responders

-

RR of all-cause mortality (responders and non-responders)

-

IDGx test

-

Annual direct medical costs (responder and non-responders

-

Annual indirect medical costs (responders and non-responders)

-

Responders

-

Non-responders

-

Model assumed a 3 year catch up period in which the SOC group response becomes equal to the IDGx group.

-

The probability of spontaneous transition between response and nonresponse was assumed to be equivalent in both arms and was thus left at zero.

Najafza deh 2017¹⁷

United States

Funding source: No funding received

Analysis: CEA

Approach: Model-based analysis; Markov model

Time horizon: 3 years (1 year, 2 years, 1o years and lifetime conducted)

Cycle length: 3 months

Perspective: US Societal

Discount rate: 3%

Analytic Approach: Univariate, one-way sensitivity analysis

Hypothetical cohort of individuals with baseline HAMD score ≥ 20 and/or a HAMA score ≥ 18³⁰, diagnosed with depression with/without anxiety

Number of participants: 10,000 individuals

Mean Age: 48 years in both groups

Sex: 27% male and 73% females in both groups

Anxiety and Depression: 65% of patients had both anxiety and depression; 35% had anxiety only

Intervention: IDgenetix (pharmaco genetic therapy)

Comparator: SOC

-

Decline in remission rates (Level 1, 2, 3 or 4)

• Level 1 treatmemt = patients receiving citalopram or an equivalent medication
• level 2–4 = non-responders and includedoptions such as bupropion, cognitive therapy, sertraline, extended-release venlafaxine, tranylcypromine, or extended-release venlafaxine plus mirtazapine

-

Decline in response rates (Level 1, 2, 3 or 4)

-

Increase in ADEs rate by treatment level (Level 1, 2, 3 or 4)

-

Suicide attempts among patients with depression

-

RR risk of suicide attempt among those who respond to treatment

-

Proportion of suicides considered successful

-

Relapse rates for remission patients (Level 1, 2, 3 or 4)

-

Relapse rates for responding patients (Level 1, 2, 3 or 4)

-

Direct costs to depression/anxiety per year

-

Indirect costs to depression/anxiety per year

-

IDgenetix costs

-

QoL for remission

-

QoL for response

-

QoL for no response

-

ADE impact on QoL

-

Assumed both groups to be identical to control for variability in outcome due to differences in populations

-

assumed that patients who responded to treatment or achieved remission could relapse, requiring new treatment

-

Assumed all patients received Treatment 1 (receiving citalopram or an equivalent medication.)

-

Assumed a 3-year time horizon for base case analysis

-

Assuming at WTP threshold of $50,000 per QALY

-

Patients were considered in remission in they if they stopped receiving treatment over two courses (6 months)

-

Assumed that relapse rates declined the longer a patient stayed in remission

-

Assumed that the relative risk of suicide attempts decreased by 0.49 if the patient responded to treatment [and assigned a rate of zero for those who achieved remission

-

Assumed that the occurrence of an ADE negatively affected a patient’s QoL and therefore penalized the QoL over that treatment period.

-

Assumed that the QoL for a patient in the remission state was identical to that of the general population

• The research questions and inclusion criteria for the review included the components of population, intervention, comparison, and outcomes.
• The review authors used a comprehensive literature search strategy.
• The review included a flow chart illustrating guideline selection and provided reasons for article exclusion
• The review authors explained their selection of all study designs for inclusion in the review.
• The included studies were described in adequate detail.
• A list of excluded studies was published.
• The sources of funding for the included studies were reported.

• It was unclear whether a published protocol was established prior to the conduct of the review.
• It was unclear whether the study selection and data extraction were performed in duplicate by two reviewers.
• The included primary studies were published in Spain, Australia, Germany, Italy and the United States and their relevance to the Canadian healthcare setting was unclear.

• The research questions and inclusion criteria for the review included the components of population, intervention, comparison, and outcomes.
• A protocol was established prior to the conduct of the review. The PROSPERO registration number was provided in the report.
• The review authors used a comprehensive literature search strategy.
• The literature search, study selection and data extraction were performed in duplicate by two reviewers.
• The included studies were described in adequate detail.
• The authors used a satisfactory technique, the Cochrane Risk of Bias Assessment Tool, for assessing the risk of bias of included individual studies
• The authors justified combining the data in a meta-analysis. The discussion and explanation of any heterogeneity observed in the results of the review was reported.
• The sources of funding for the included studies were reported.
• The potential sources of conflict of interest and funding were disclosed.

• A list of excluded studies was not published.
• The review authors did not explain their selection of RCT study design for inclusion in the review.

• The research questions and inclusion criteria for the review included the components of population, intervention, comparison, and outcomes.
• The review authors used a comprehensive literature search strategy.
• The review authors explained their selection of all study designs for inclusion in the review.
• The included studies were described in adequate detail.
• The sources of funding for the included studies were reported.
• The discussion and explanation of any heterogeneity observed in the results of the review was reported.
• The potential sources of conflict of interest and funding were disclosed.

• A protocol was established but not registered prior to the conduct of the review.
• The study selection and data extraction were not reported to be performed in duplicate by two reviewers.
• A list of excluded studies was not published.
• An investigation of publication bias and its impact on the results of the review were not reported.

• The objective, main outcomes, characteristics, interventions, confounders and main findings of the study were clearly described.
• The patient adherence to with the interventions were likely reliable due to observed dosing of medication in hospital. The patients were randomized to the treatment groups using a random number list.
• The patients in different intervention groups were recruited from the same population and over the same time period.
• The statistical tests used to assess the main outcomes were appropriate.
• The main outcome measures used were reliable as it was depression severity measured as a validated HAMD-17 score.
• Probability values were reported as p values for the main outcomes.
• Potential conflicts of interest were reported in the article.

• The study was single-blinded, with the blinded group being the clinical psychiatrist.
• It was unclear whether the patients who participated, staff, places, and facilities in the study in China were representative of the Canadian population.
• The patients who were asked to participate and prepared to participate in the study were recruited via convenience sample (patients presenting to the hospital). It was unclear whether they were representative of the entire population of adult patients with depression.
• A power calculation was not conducted to determine if the sample was of an adequate size for clinical significance.

• The research question, economic importance of the research question, and rationale for choosing alternative interventions compared were clearly stated
• The viewpoint/perspective of the analysis were clearly stated.
• The treatment strategies being compared were clearly described
• The form of economic evaluation used was stated
• The choice of form of economic evaluation was justified in relation to the questions addressed

• The sources of effectiveness estimates and screening/treatment costs were provided
• The design and results of effectiveness studies from which assumptions were drawn were provided although not clear as some points.
• The primary outcome measures for the economic evaluation were clearly stated
• Methods to value benefits were stated
• Details of the subjects from whom valuations were obtained were given
• Methods for the estimation of unit costs were described
• The structure of the model was clearly described using figures

• Time horizon of costs and benefits was stated (12 weeks)
• Discount rates, choices of discount rates, and explanation of costs/benefits not being discounted were provided
• The approach to sensitivity analysis was given
• The choice of variables for the sensitivity analysis were justified
• Major outcomes were presented in a disaggregated form
• The answer to the study question was given
• Incremental analysis was reported
• Conclusions follow from the data reported
• Conclusions were accompanied by appropriate caveats

• The authors declared that they had no potential conflicts of interest
• Sources of funding were disclosed and were unlikely to have had an effect on the findings of the study

• The viewpoint and perspective were not justified.
• The relevance of productivity changes was not discussed
• Details of currency adjustments or conversions were not provided
• Evaluation is a Netherlands-based study and may not be generalizable to other health systems
• A societal perspective may not be appropriate in the Canadian context
• The time horizon may be considered short although the authors did describe their reasoning for this.
• There is some uncertainty with key parameters in the model which may affect the ICER.
• One study was the source of model parameters and studied one antidepressant treatment (venlafaxine; assumptions were made for other antidepressants), which may overestimate the effect and results of this study
• Multiple assumptions made in the study may affect the results and interpretation of the study.

• The viewpoint/perspective of the analysis was stated
• The research question, economic importance of the research question, and rationale for choosing alternative interventions compared were clearly stated
• The treatment strategies being compared were clearly described
• The design and results of effectiveness studies from which assumptions were drawn were not provided
• The form of economic evaluation used was stated
• The choice of form of economic evaluation was justified in relation to the questions addressed

• The sources of effectiveness estimates and screening/treatment costs were provided
• The primary outcome measures for the economic evaluation were clearly stated
• Methods to value benefits were stated
• Details of the subjects from whom valuations were obtained were given
• Methods for the estimation of unit costs were described
• The structure of the model was clearly described using figures

• Time horizon of costs and benefits was stated (12 weeks)
• Discount rates, choices of discount rates, and explanation of costs/benefits not being discounted were provided
• The approach to sensitivity analysis was given
• The choice of variables for the sensitivity analysis were justified
• Major outcomes were presented in a disaggregated form
• The answer to the study question was given
• Incremental analysis was reported
• Conclusions follow from the data reported
• Conclusions were accompanied by appropriate caveat

• The author disclosed a potential conflict of interest as being consultant for the manufacturer of IDGx
• Sources of funding were disclosed and may have had an effect on the findings of the study
• The viewpoint/perspective of the analysis was not justified although the authors noted that there may be some uncertainty with this.
• Data input were taken from single references, rather than a synthesis or meta-analysis of estimates from multiple sources
• The relevance of productivity changes was not discussed
• Details of currency adjustments or conversions were not provided (assumed to be US dollars)
• The data and studies used as model parameters were based on different populations and studies
• Data and studies are based on a 12-week follow-up periods which may inaccurately depict the patient’s response to treatment
• There may be slights over-estimation of mortality rates for depressions, since suicide rates were derived from a study focusing on treatment-resistant MDD
• The findings of this US based evaluation may not be generalizable to other health systems
• A societal perspective may not be appropriate in the Canadian context
• There was some uncertainty with key parameters in the model which may affect the ICER.
• Utility scores may be based on data from samples that were different from the populations that they were applied affecting the overall results of the model.
• There was uncertainty regarding the length of the follow-up of patients and treatment.

• The viewpoint/perspective of the analysis was stated while supplemental time horizons were used for another analysis (12 weeks, 1, 2, 10 years and lifetime)
• The research question, economic importance of the research question, and rationale for choosing alternative interventions compared were clearly stated
• The treatment strategies being compared were clearly described
• The design and results of effectiveness studies from which assumptions were drawn were not provided
• The form of economic evaluation used was stated
• The choice of form of economic evaluation was justified in relation to the questions addressed

• The sources of effectiveness estimates and screening/treatment costs were provided
• The primary outcome measures for the economic evaluation were clearly stated
• Methods to value benefits were stated
• Details of the subjects from whom valuations were obtained were given
• Methods for the estimation of unit costs were described
• The structure of the model was clearly described using figures

• Time horizon of costs and benefits was stated (12 weeks)
• Discount rates, choices of discount rates, and explanation of costs/benefits not being discounted were provided
• The approach to sensitivity analysis was given
• The choice of variables for the sensitivity analysis were justified
• Major outcomes were presented in a disaggregated form
• The answer to the study question was given
• Incremental analysis was reported
• Conclusions follow from the data reported
• Conclusions were accompanied by appropriate caveats

• Sources of funding were disclosed

• The author disclosed a potential conflict of interest as a consultant for the manufacturer of IDGenetix
• The viewpoint/perspective of the analysis was not justified
• Data input were taken from single references, rather than a synthesis or meta-analysis of estimates from multiple sources
• The relevance of productivity changes was not discussed
• Details of currency adjustments or conversions were not provided (assumed to be US dollars)
• The findings of this US based evaluation may not be generalizable to other health systems
• A societal perspective may not be appropriate in the Canadian context
• There was some uncertainty with key parameters in the model which may affect the ICER.
• Results of the study did not apply to patients with mild depression with/without anxiety.
• The analysis used the same cost estimates for each severity level of depression.

Key Question #1b Impact of pharmacogenomics testing on patient outcomes for depressive disorders

Remission:

Winner, 2013²² (guided - GeneSight PGx test panel n = 26 vs unguided n = 25)

• At 10 weeks, patients achieved remission (Ham-D17 <7):

  • 20% of guided patients vs, 8.3% of unguided (OR=2.75; 95% CI, 0.48–15.8; P = NS).

Singh, 2015²³ (guided - CNSDose assay n = 74 vs unguided n = 74)

• At 12 weeks, guided patients more often obtained remission (HAM-D17 <7)

  • OR=2.52; 95% CI, 1.71–3.73; P < 0.0001
• Number needed to test for remission = 3 (95% CI, 1.7–3.5).

Hall-Flavin, 2013²¹ (guided - GeneSight n = 114 vs unguided n = 113)

• At 8 weeks, more guided patients obtained remission (QIDS-C16<6) compared with unguided patients (OR=2.42; 95% CI, 1.09–5.39; P = 0.03).
• HAM-D17 and PHQ-9 results not significantly different except for results using data imputation to account for 27% lost to follow-up.

Breitenstein, 2014²⁴ (guided - ABCB1 test n = 58 vs unguided n = 58)

• guided patients more often in remission (HAM-D21 <10) at treatment week 4 compared with unguided patients (83.6% vs 62.1%; P = 0.005). HAM-D21 at admission >14.

  • Required change in score may not be clinically relevant.

Response to treatment:

Winner, 2013²² (guided - GeneSight n = 26 vs unguided n = 25, all genotyped)

• At 10 weeks, 36% of guided patients responded (>50% reduction in HAM-D17) vs 20.8% of unguided patients (OR=2.14; 95% CI, 0.59–7.69; P = NS).

Hall-Flavin, 2013²¹ (guided - GeneSight n = 114 vs unguided n = 113, all genotyped)

• At 8 weeks more guided patients responded (>50% reduction in score from baseline) vs unguided patients as measured by:

  • QIDS-C16 (OR=2.58; 95% CI, 1.33–5.03; P = 0.005)
  • HAM-D17 (OR=2.06; 95% CI, 1.07–3.95; P = 0.03)
  • PHQ-9 (OR=2.27; 95% CI 1.20–4.30; P = 0.01)
  • Results using data imputation to account for 27% loss to follow-up were statistically significant except for QIDS-C16.

Hall-Flavin, 2012²⁰ (guided - GeneSight n = 25 vs unguided n = 26; all genotyped)

• 8-wk score reductions:

  • QIDS-C16: 31.2% for guided patients vs 7.2% for controls (P = 0.002).
  • HAM-D17: 30.8% for guided patients vs 18.2% for controls (P = 0.04).

Rundell, 2011²⁵ (guided n = 29 vs unguided n = 17) – Intervention is At least one of CYP2D6, CYP2C19, CYP2C9, and/or serotonin transporter genotype 5-HTTLPR vs. no test ordered

• CYP450 categories: No significant differences in serial PHQ-9 scores over time.
• 5-HTTLPR categories: L/L genotype patients had greater PHQ-9 score improvement than other genotypes at times 4 and 5 (P = 0.02 to P = 0.05).
• Adjusted post-day 14 PHQ-9 scale slopes and differences in pre- to post-baseline scale slopes were not significantly different among genotype categories.

Adherence, tolerance, adverse events:

Singh, 2015²³ (guided - CNSDose n = 74 vs unguided n = 74)

• Unguided patients were less able to tolerate medications, requiring dose reduction or cessation (OR=1.13; 95% CI, 1.01–1.25; P = 0.0272).
• guided patients took sick leave less often (4% vs 15%; P = 0.0272) and of less duration when needed (4.3 vs 7.7 days; P = 0.014).

Hospital stay/Healthcare utilization:

Breitenstein, 2014²⁴ (guided - ABCB1 test n = 58 vs unguided n = 58)

• Dose increases in genotype-appropriate antidepressants were associated with shorter hospital stays (P = 0.009). Hospital stay for patients with unfavorable ABCB1 genotype was reduced by 4.7 weeks if dose was increased more than 1.5-fold.

Key Question #4: What are the costs and cost-effectiveness of genetic testing to guide the selection or dose of medications?

Cost-effectiveness studies:

Perlis, 2009²⁶ HTR2A PGx testing either before first-line tx (Test 1st) or after first-line tx failure (Test 2nd) vs no testing (Ctl):

• Direct medical costs including outpatient and inpatient treatment, meds
• Test 1st + bupropion tx for test-negative patients ↑ cost by $505/pt but provided 0.0054 QALY for ICER of $93,520/QALY; therefore, not cost-effective.

Olgiati, 2012²⁷, 5-HTTLPR PGx testing vs none in high income Western European countries:

• Estimated costs of meds, outpatient and inpatient care, and genetic testing in Western European healthcare systems
• Incremental benefit of PGx 0.062 QALWs for response + 0.016 QALWs for side effects
• Overall incremental PGx benefit 0.156 QALWs
• Estimated overall cost of healthcare Intl $2,242 (PGx) vs Intl $2,063 (Unguided)
• Incremental cost of PGx testing was Intl $179 and the ICER was Intl.$1,147

Cost-utility studies:

Herbild, 2009²⁸ CYP2D6 PGx testing vs none, willingness-to-pay for PGx:

• Willingness to pay for a 10% probability of 1 antidepressant change or for the reduction of 1 month of time for dosage adjustments exceeded test cost in Denmark

Relevant practice guidelines

• World Federation of Societies for Biological Psychiatry guideline, 2013³² recommended: “In possibly nonadherent patients (e.g., low drug plasma levels despite high doses of the antidepressant), a combination of TDM and genotyping may be informative. Such analyses can aid in identifying those individuals who are slow or rapid metabolizers of certain antidepressants.” (p.22)¹³

  • Quality: 5.0 – Fair
  • Limitations: Search terms and dates literature covered NR; criteria for selecting evidence and how the body of evidence was evaluated for bias NR

Clinical effectiveness:

Remission

“In all studies, the direction of results suggests that genotyped pts are more likely to obtain remission. But results are not consistently statistically significant and in 1 study may not be clinically relevant.” (p.15)¹³

Response to treatment:

“Results are in the direction of improved response for genotyped patients. Only 1 study used defined measures of response and obtained statistically significant results.” (p. 15–16)¹³

Adherence, tolerance, adverse events:

“In 2 of 3 studies, results indicate increased tolerance of medications when prescribed with knowledge of PGx results.” (p. 16)¹³

Hospital stay/Healthcare utilization:

“Results indicate PGx for ABCB1 variants may result in better anti-depressant dosing and shorter hospital stays; not generalizable” (p. 17)¹³

Cost effectiveness:

“One study found PGx testing not to be cost-effective; 1 modeling study of a hypothetical pt cohort estimated an increased overall cost of healthcare with PGx vs Ctl for an incremental benefit in QALW.” (p.20) ¹³

Cost utility:

“Utility increases with decreases in the number of changes in meds or ↓ times for dosage adjustments.” (p.20) ¹³

Guideline recommendation:

“In possibly nonadherent patients (e.g., low drug plasma levels despite high doses of the antidepressant), a combination of TDM and genotyping may be informative. Such analyses can aid in identifying those individuals who are slow or rapid metabolizers of certain antidepressants.” (p.22)¹³

“In summary, the evidence base for pharmacogenomic testing for the psychiatric disorders of interest for this report is extremely limited and compromised and is considered to be of low to very low quality, depending on the outcome measured. As such, the evidence is insufficient for forming conclusions regarding clinical use.” (p.56)¹³

Relevant primary studies:

Bradley, 2018³⁰

• Intervention vs. comparator: pharmacogenetic (NeuroIDgenetix)-guided (n = 352) vs unguided (n = 333)
• Remission:

  • Guided group, n/N: 14/40,
  • Unguided group, n/N: 7/53,
  • RR, 95%CI: 2.65, (1.18–5.95)

Greden, 2018³¹

• Intervention vs. comparator: pharmacogenetic (GeneSight)-guided (n = 681) vs unguided (n = 717):
• Remission:

  • Guided group, n/N: 93/607,
  • Unguided group, n/N: 57/560,
  • RR, 95%CI: 1.51, (1.11–2.05)

Perez, 2017²⁹

• Intervention vs. comparator: pharmacogenetic (Neuropharmagen)-guided (n = 155) vs unguided (n = 161)
• Remission:

  • Guided group, n/N: 48/141,
  • Unguided group, n/N: 46/139,
  • RR, 95%CI: 1.03, (0.74–1.43)

Singh, 2015²³ – reported in Washington State, 2016¹³, analyzed in pooled analysis

• Intervention vs. comparator: pharmacogenetic (CNSDose)-guided (n = 74) vs unguided (n = 74)
• Remission:

  • Guided group, n/N: 52/74,
  • Unguided group, n/N: 21/74,
  • RR, 95%CI: 2.52, (1.71–3.73)

Winner, 2013²² – reported in Washington State, 2016¹³, analyzed in pooled analysis

• Intervention vs. comparator: pharmacogenetic-guided (n = 25) vs unguided (n = 24)
• Remission:

  • Guided group, n/N: 5/25,
  • Unguided group, n/N: 2/24,
  • RR, 95%CI: 2.40, (0.51–11.21)

“Our meta-analysis showed pharmacogenetic-guided prescribing has a positive effect on the likelihood of achieving symptom remission. However, inclusion criteria of the included studies suggest this positive effect on remission may be confined to individuals with moderate to severe depression and a history of inadequate response or intolerability to previous psychotropic medications.” (p. 43)⁶

“Our updated systematic review and meta-analysis suggests pharmacogenetic-guided DST treatment is superior to treatment as usual in relation to remission likelihood, specifically among those with inadequate response or intolerability to previous psychotropic medications and perhaps more noticeably among individuals with more severe depressive symptoms. Thus, the results to date, suggest pharmacogenetic-guided DSTs merit consideration by clinicians treating patients who have not responded or have not been able to tolerate one or more psychotropic medications.” (p. 43–44)⁶

“We systematically identified and assessed five RCTs that examined the effect of pharmacogenetic-guided antidepressant prescribing on remission in the management of MDD. Our meta-analysis showed pharmacogenetic-guided prescribing has a positive effect on the likelihood of achieving symptom remission. However, inclusion criteria of the included studies suggest this positive effect on remission may be confined to individuals with moderate to severe depression and a history of inadequate response or intolerability to previous psychotropic medications.” (p. 43)⁶

Clinical effectiveness:

Hall-Flavin, 2012²⁰ – reported in Washington State, 2016¹³, data not extracted here

Hall-flavin, 2013²¹ – reported in Washington State, 2016¹³, data not extracted here

Winner, 2013²² – reported in Bousman, 2019⁶ and Washington State, 2016¹³, data not extracted here

Singh, 2015²³ – reported in Bousman, 2019⁶ and Washington State, 2016¹³, data not extracted here

Cost-effectiveness:

Hornberger, 2015³³

Use model from Perlis 2009 for cost-effectiveness analysis

Study data from Hall-Flavin 2012, Hall-flavin 2013, and Winner 2013

Estimated change in QALYs: increase by 0.316 years for PGx guided therapy

Projected savings:

• Saving in direct medical costs: $3,711
• Saving in work productivity costs per patient over the lifetime: $2,553

Probability of GeneSight testing being cost-effective at the WTP threshold of $50,000: 94.5%.

Winner, 2015³⁵ – reported in Washington State, 2016¹³, data not extracted here

Winner, 2013²² – reported in Bousman, 2019⁶ and Washington State, 2016¹³, data not extracted here

Olgiati, 2012²⁷ – reported in Washington State, 2016¹³, data not extracted here

Perlis, 2009²⁶ – reported in Washington State, 2016¹³, data not extracted here

Conclusion on Hornberger, 2015 “Therefore, their results suggested that combinational pharmacogenomic testing could be a cost-effective intervention. Notably, however, their projections were based mostly on studies of poor quality; lacking appropriate randomization and blinding; to determine efficacy of GeneSight testing, 93.3% of the pooled results was based on 2 open-label, nonrandomized studies, while only 6.7% of their pooled results was based on a randomized, controlled, and double-blinded study. Therefore, the pooled efficacy (pooled effect of testing on response rate calculated to be 1.71 [95% CI 1.17 2.49]) was based mostly on low-quality studies. Since the model of cost-effectiveness is heavily weighted on intervention efficacy effect size, the validity of the results of this analysis is questionable as the reliability of the calculated efficacy may be poor.”(p.727) ¹⁴s

“In conclusion, currently available evidence for improved clinical outcomes from pharmacogenomic testing is limited. Clinical trials suggestive of a positive effect of pharmacogenomic testing on clinical outcomes in MDD were mostly of low quality, often lacking randomization and blinding, and were vulnerable to bias from industry funding. Further, results from a randomized, double-blind clinical trial of GeneSight did not reach statistical significance; however, notably, they may have been underpowered. One randomized, double-blind clinical trial of CNSDose found a statistically significant increase in remission rates, but this has yet to be independently replicated.29 Taken together, results from these studies suggest that. further studies are required and merited to determine The impact of these tests on clinical outcomes, namely in the rate (ie, time to improvement) and amount (ie, response and remission rates) of therapeutic improvement. Well-designed clinical trials with adequate sample sizes) randomization, and blinding are required prior to the routine implementation of pharmacogenomic testing into clinical practice. If testing is found to improve clinical outcomes, the cost-effectiveness of testing should also be further evaluated based on the results from high-quality studies.” (p.728) ¹⁴

• Significantly decreased from baseline to 8 weeks within the guided and unguided groups (P < 0.01)
• No significant difference was found in the HAMD-17 total scores at each time point between the unguided and guided groups

  • At 2 weeks, P = 0.696
  • At 4 weeks, P = 0.901
  • At 8 weeks, P = 0.205
• The reduction ratio of HAMD-17 scores at 8 weeks: 60.86% in the guided group vs. 52.38% in the unguided group with no significant difference (P = 0.210)

  • Reduction ratio = (HAMD-17total_1-HAMD-17total_2)/HAMD-17total_1.
  • HAMD-17total_1 refers to the HAMD-17 total scores at baseline
  • HAMD-17total_2 is the HAMD-17 total scores after 8 weeks of treatment

Response rate after 8 weeks of treatment:

• Guided group 74.19% (23/31) vs unguided group 57.5% (23/40), not statistically significantly different (P = 0.144)

Remission rate:

Guided group 61.29% (19/31) vs. unguided groups 45.0% (18/40) (P = 0.173)

Not statistically significantly different

Anxiety symptoms measured by HAMA total scores:

No significant difference observed at each time point between two groups

• At 2 weeks, P = 0.985
• At 4 weeks, P = 0.889
• At 8 weeks, P = 0.961

Routine blood test, liver function, renal function, and electrocardiogram examinations:

After 8 weeks of treatment, no abnormalities were found

Adverse reactions and medication tolerability measured by TESS:

Incidence rate of adverse reactions:

Guided group 55.56% vs. Unguided group 57.89%

No statistical difference between the two groups.

Frequency of adverse reactions:

Guided groups 14 cases vs. Unguided groups 19

Cases, respectively.

Main adverse reactions:

Headache, dizziness, drowsiness, nausea, vomiting, dry mouth, constipation, diarrhea, decreased appetite, and tachycardia

• Frequency NR

Costs:

PGx = €970 (CI: €799 − €1244)

No PGx = €881 (CI: €723 − €1130)

Incremental cost:

€89 (CI: €39 − €152)

QALYs:

PGx = 0.146 (CI: 0.133 − 0.159)

No PGx = 0.145 (CI: 0.132 − 0.157)

Incremental QALY :

0.001 (CI: 0.001 − 0.002)

ICER :

€77,406 per QALY

48% of the simulations were below the WTP threshold of €80,000 per QALY

Scenario Analysis:

-

Most sensitive to a delayed effect of a week for PGx; not cost-effective

-

Most sensitive to productivity losses were not considered, the ICER is not cost-effective

“According to our model, we cannot unequivocally conclude that screening for CYP2D6 in primary care patients using antidepressants is be cost-effective, as the results are surrounded by large uncertainty. Therefore, information from ongoing studies should be used to reduce these uncertainties.” (pg.1)

“Although patients’ genetic properties could be taken into account, other diagnostics, such as therapeutic drug monitoring, psychosocial factors, etc., remain necessary to optimise treatment as ‘PGx’ still misses a part of the patients with a poor metabolizer status or that are still not responding to treatment.” (pg.9)

Total Costs:

Moderate to Severe Depression

SOC = $47,295

IDGx = $44, 697

Severe Depression

SOC = $47,025

IDGx = $41,215

Incremental cost:

Moderate to Severe Depression

$-2,598

Severe Depression-$5,810

QALYs:

Moderate to Severe DepressionSOC = 1.97

IDGx = 2.07

Severe DepressionSOC = 0.356

IDGx = 0.311

Incremental QALY:

Moderate to Severe Depression= 0.01

Severe Depression= 0.17

ICER:

IDGx dominates over SOC

Sensitivity Analysis:

-

Model most sensitive to annual health care costs for responders

-

Model most sensitive to annual health care costs for non-responders

“The IDGx-guided treatment producing both QALYs gained and cost savings, the treatment “dominates” the SOC treatment.” (pg.5)

“Given the increased need for a variety of health care providers to prescribe and manage antidepressants, pharmacogenetic tests are a valuable tool that demonstrate improved patient outcomes in real-world settings and are strongly positioned to help reduce the economic burden of depression.” (pg.7)

Total Costs:

TAU = $14,659 (CI: S10,384 – 19,275)

IDgenetix = $14,124 (CI: $10,703 – 17,630)

Incremental cost:

-$535

QALYs:

IDgenetix = 2.09 (CI: 1.88 – 2.88)

TAU = 1.94 (CI: 1.66 – 2.21)

Incremental QALY:

0.15 (0.04 – 0.28)

ICER:

IDGenetix dominates over TAU

The probability of IDGenetix being cost-effective over TAU is 90%

Sub-group Analysis:

Patients with/without anxiety = $35/QALY

Patients with severe depression and/or severe anxiety = IDGenetix dominates of TAU

Sensitivity Analysis:

• Model most sensitive most to assumption about remission and response rates of alternative treatment strategies

“In summary, we found that implementing IDgenetix guided treatment of patients with moderate-to-severe depression and/or anxiety is likely to result in cost savings and improved QOL, compared with TAU…. Overall, several features of the IDgenetix test, such as the efficacy of a guided treatment strategy, one-time test cost, and the prospect of using test results for guiding future episodes of depression or anxiety, make it a potentially dominant strategy compared with usual care.” (pg.12)