We designed a cross-sectional observational study, between August 2018 and April 2019, among the participants of the DRIVE project (NTC03526939). This project had the primary objective to evaluate the impact of a community-based intervention for HIV screening and treatment among PWIDs to control HIV transmission in Hai Phong, Vietnam and consisted of 4 annual respondent-driven sampling surveys and biannual visits for those enrolled in the cohorts (39). PWIDs were eligible if they were (i) older than 18 years old, (ii) able to provide a signed informed consent, and (iii) actively injecting drugs as assessed by visible marks of injection and a urinary test positive for heroin and/or methamphetamine. In a face-to-face interview, community workers gathered information on socio-demographic data and drug use. Nurses provided HIV and hepatitis C virus (HCV) tests. During the biannual visit, participants had a consultation with a psychiatrist who diagnosed potential mental health problems, using the Mini International Neuropsychiatric Interview (MINI), a structured psychiatric interview based on the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV) criteria (40). If necessary, psychiatric treatments were initiated. The MINI test was used to diagnose current and/or lifetime MDE, current and/or lifetime PS, and to evaluate suicidal risk.

For the present study, we included PWIDs who had banked blood samples and who had a psychiatric interview (Supplementary Figure 1). PWIDs were next divided into two groups: a group of PWIDs with psychiatric conditions, defined by subjects with at least one psychiatric condition diagnosed among either MDE or PS, and a control group of PWIDs without either of these conditions. The population with MDE included both those with a current or a lifetime MDE. Similarly, the population with PS included both those with a current or a lifetime PS.

The present study complies with the Declaration of Helsinki and was approved by the Institutional Review Board of the Haiphong University of Medicine and Pharmacy, Vietnam (#01/HPUMPRB) and Mount Sinai Beth Israel hospital, New York, United States.

In total, 80 μl of whole blood was collected on dried blood spot cards (DBS, Whatman™ 903, GE Healthcare Bio Sciences Corp.), dried, and stored at −80°C. DNA was extracted after blood elution using a QIAamp DNA Mini Kit, following the instructions of the manufacturer, and stored at −80°C until further process.

Mitochondrial DNA copy number was assessed by a commercial assay (QuickScanTM Mitox Kit, Primagen, Amsterdam, Netherlands) targeting rRNA 16S in the mtDNA genome and snRNP U1A in the genomic DNA. Briefly, 2.5 μl of the sample extract was added to a mixture of primers (Supplementary Table 1) at a final concentration of 500 nM and LC480^® SYBR Green I Master Mix 2X (Roche, Germany) for a final volume of 25 μl. Genomic and mitochondrial targets were amplified separately in LightCycler^® 480 Instrument (Roche, Germany). The amplification program consisted of 5 min of incubation at 95°C followed by 45 cycles of 10 s at 95°C, 20 s at 60°C, and 20 s at 72°C. MCN was obtained by reporting the Ct of each target on an external standard curve, made with serial dilution of calibrator consisting of plasmids with rRNA 16S and snRNP U1A inserts. In each plate, one calibrator, one control, and one no template control were added. The plate was valid if the calibrator and control were within the expected range and no amplification occurred in the no template well. The sample was valid if the Ct value did not exceed 30. MCN was expressed in copies of mtDNA/cell (c/cell), assuming each cell has two copies of genomic DNA.

Mitochondrial DNA deletion was quantified by quantitative PCR (qPCR) that targeted two regions of the mtDNA, the ND1 gene and the ND4 gene (primers described in Supplementary Table 1); in the latter, more than 85% of the deletions reported so far were found (41). The reaction mixture consisted of 5 μl of sample extract, forward primers, and probes at a final concentration of 100 nM, reverse primers at a concentration of 250 nM, and LC480^® Probes Master Mix 2X for a final volume of 20 μl. The amplification program consisted of 5 min of incubation at 95°C, 45 cycles of 10 s at 95°C, 30 s at 60°C, and 1 s at 72°C, followed by a cooling step of 10 s at 40°C. Each plate contained one DNA extract from plasma considered as a control and a no template control. Validation of the amplification followed the sample rules for MCN. The deletion rate was obtained using the 2^–ΔΔCt method (42), and the result was normalized to a 10 ng DNA extract of plasma-rich platelets.

Telomere length was measured by a real-time PCR, using a technique described elsewhere (43). Succinctly, 10 ng of DNA samples was added to a mix of primers (final concentration of 900 nM, Supplementary Table 1) and LC480^® SYBR Green I Master Mix 2X (Roche, Germany) for a final volume of 30 μl. The following programs were applied: 10 min at 95°C, 40 cycles of amplification of 15 s at 95°C, 30 s at 60°C, and 30 s at 72°C. Each plate contained standards for telomere sequence ranging from 1.18 × 10⁸ to 1.18 × 10⁴ kb and for genomic sequence targeting, a single copy gene (namely, 36B4) ranging from 2.63 × 10⁶ to 2.63 × 10² diploid genome copies. They also contained HEK293T DNA extracts and a well with no extract for the validation of sample amplification (44).

Excessive alcohol consumption was determined by the Alcohol Use Disorders Identification Test-Consumption (AUDIT-C) questionnaire, with a threshold greater than or equal to 4 for men and 3 for women (45). Poly substance use included any excessive alcohol consumption, daily heroin injection, or frequency of methamphetamine consumption greater than or equal to 4 times a month (38). Subjects were considered infected with HIV if HIV serology was positive and infected with HCV if HCV serology and HCV-viral load were positive. At the time of the sampling, most of the HIV-positive subjects were treated with antiretrovirals consisting of a combination of tenofovir/lamivudine/efavirenz (TDF/3TC/EFV) for the HIV infection, but none was treated for HCV.

Population characteristics are given as frequencies and percentages for categorical variables or mean values with standard deviation (SD) for numerical variables. A chi-squared test or t-test was performed for group comparisons (α = 5%). Raw data of TL/MCN/MDD were given with a median and confidence interval of 95% (CI 95%). For group comparison, TL, MCN, and MDD values were first log-transformed and then adjusted for age using linear regression. The difference between the group was assessed using the Wilcoxon signed post hoc test for the Tukey-adjusted least square mean (LSM) rank test (α = 5%).

In multivariate analysis, associations between psychiatric conditions (MDE and PS) and age-related biomarkers were assessed by binomial logistic regression, using MCN, MDD, and TL as continuous variables and adjusted for possible confounding factors (i.e., factors associated with both psychiatric disorder and aging-related biomarkers). Therefore, the selection of the variables was based on a literature review rather than a statistical criterion. These factors were related to heroin, methamphetamine, and alcohol consumption, which are risk factors for psychiatric disorders and have shown their potency to accelerate aging (37, 46, 47). We also tested HIV and HCV infections as they have been both linked with neuropsychiatric outcomes and accelerated aging (46, 48, 49). We also included ARV and methadone treatments, because they are able to induce immuno-inflammatory anomalies, which can accelerate premature cellular aging (50, 51) and direct inhibitory effect on enzymatic processes. At last, we included age and sex, which have both been linked with TL and mitochondrial dysfunctions: TLs are shortened with age and are longer in women than in men (52); mitochondrial dysfunction is accumulated with age (53). Altogether, the following variables were used: “age” (continuous), “sex” (men/women), “daily consumption of heroin” (yes/no), “frequency of methamphetamine consumption per month” (<4 times a month/ ≥4 times a month), “excessive alcohol consumption” (yes/no), “HIV serology” (positive/negative), “HCV infection” (positive/negative), “declare being on methadone assisted therapy” (yes/no), and “under ARV therapy” (yes/no). The number of variables included in each model followed the “one in ten” rule. No multiple testing correction was done. Final models were selected in a forward stepwise manner, using the smallest Akaike Information Criterion. Lastly, the models’ goodness of fit was assessed with Homer and Lemeshow test (α = 5%). The odds for any psychiatric condition depending on TL (continuous variable) were given for 0.1 unit change of TL, i.e., 0.1 log₁₀ (kb/cell). Statistical analyses were performed on SAS^® studio university (Copyright© 2012–2020, SAS Institute Inc., Cary, NC, United States).

Overall, 299 subjects had a consultation with a psychiatrist between August 2018 and April 2019. In total, twenty subjects were excluded from the analysis because of incomplete data, and thirteen were also excluded because they had an isolated suicidal risk. In total, 130 subjects were diagnosed with at least one psychiatric condition and 136 had no diagnosis of psychiatric condition and were defined as our control population. A patient flowchart is available in Supplementary Figure 1. Out of the 130 PWIDs with a psychiatric condition, 110 (84.6%) were diagnosed with MDE, 74 (56.9%) with PS, and 54 (41.5%) were diagnosed with both (Table 1).

People who inject drugs with any psychiatric condition differed from PWIDs with no psychiatric condition. Notably, they were older, injected heroin for longer, smoked methamphetamine more often, used street methadone more frequently, had excessive alcohol consumption more often, and were less committed to their methadone program. However, the use of other drugs (cannabis, cocaine, amphetamine, ecstasy, and ketamine) was very limited and did not differ significantly between the two groups (data not shown). The prevalence rates of viral infections between the two groups were similar; more than two-thirds of HIV-infected participants were receiving first-line ARV treatment (Table 2).

In total, 224, 256, and 253 samples had validated data for MCN, MDD, and TL, respectively. Reasons to invalidate the data were unmet quality control criteria and no blood sample available. TL was shorter after adjustment for age in PWIDs with an MDE (149.4 [139.9; 164.9]) and in PWIDs with both MDE and PS (154.2 [129.9; 162.7]), when compared to the control population. These differences were not significant (Table 3). TL was longer among PWIDs with PS (211.4 [147.9; 230.4] vs. 170.3 [160.8; 181.4], p = 0.01; Table 3). MCN values did not statistically differ between groups of PWID with MDE, PS, or both and the control group (Table 3). Finally, rate of MDD was higher in PWID with PS (0.55 [0.47; 0.69]) when compared to control group (0.72 [0.64; 0.80], but not statistically significant after adjustment for age, Table 3).

Additionally, PWID diagnosed with at least two episodes of MDE in their lifetime had shorter TL when compared to control group PWIDs after adjusting for age (p = 0.0025, respectively, Table 4). MCN and rate of MDD were not different according to the number of depression episodes (Table 4).

Because MDE and PS are two nosological different conditions, we grouped participants with MDE and PS regardless of their status for the other psychiatric conditions. Neither MCN nor MDD was associated with any psychiatric disorder in multivariate analyses when adjusted for confounders (data not shown). Regarding TL, we first found a strong interaction between psychiatric disorders, TL, and HCV infection. Therefore, we performed two analyses stratified on the subject of HCV status.

After stratification, the association between MDE and TL was strong for HCV-infected PWIDs, after adjustment for age, sex, daily heroin consumption, alcohol consumption, and not being on methadone treatment (adjusted odds ratio (aOR): 0.53 [0.36; 0.80]). Excessive alcohol consumption and not receiving methadone were associated risk factors of MDE (aOR: 3.23 [1.18; 8.87] and 3.03 [1.18; 7.80], respectively). Similar associations between PS and TL or PS and excessive alcohol consumption were observed among HCV-infected PWIDs, with aOR of 0.59 [0.39; 0.88] and 2.77 [1.37; 8.81], respectively, after adjustment for age, sex, and methadone treatment (Tables 5, 6). Among HCV-uninfected PWIDs, no significant association was observed between MDE and TL (aOR: 1.05 [0.72; 1.52]), after adjustment for age, methamphetamine consumption, and methadone treatment. PS was, however, associated with longer TL (aOR: 1.60 [1.01; 2.53]) and HIV-positive serology (aOR: 4.42 [1.20; 16.27]), after adjustment for age (Tables 5, 6).