Sixty Korean adults who visited the fatigue clinic of Gangnam Severance Hospital, Yonsei University College of Medicine with moderate to severe fatigue for longer than 6 months were recruited between July and November 2013. All participants were assessed for eligibility using the following criteria: age ≥20 years, Brief Fatigue Inventory-Korean version (BFI-K) score ≥4 (moderate to severe fatigue), no pregnancy or lactation, no medical history of malignancy, and no acute or chronic severe illness (history of common cold, acute gastroenteritis, uncontrolled diabetes, uncontrolled hypertension, liver or renal disease, gout, or renal calculi). Participants were excluded for oral vitamin supplement intake, abnormal laboratory tests (white blood cell count, hemoglobin, creatinine, serum glutamic oxaloacetic transaminase, serum glutamic pyruvic transaminase, thyroid stimulating hormone [TSH]), or urinalysis in the preceding 3 months.

The institutional review board of Gangnam Severance Hospital, Yonsei University College of Medicine, approved this study, and written informed consent was obtained from each participant (#3-2012-154). The original clinical trial is registered at ClinicalTrials.gov (NCT01926132) and was permitted by the Korean Ministry of Food and Drug Safety (20130091445).

We measured height, weight, and blood pressure. Weight (kg) and height (cm) were measured using standardized scales and stadiometers while wearing light clothing. Body mass index (BMI) was calculated as body weight divided by the height squared (kg/m²). All physical measurements were made twice and analyses were performed on averaged values. All participants answered questionnaires about comorbidities, medication history, smoking status (never smoked, ex-smoker, or current smoker), alcohol intake (never drink, or more than once per week), physical exercise of at least 30 min/d three times per week, and coffee consumption (per week).

We measured fatigue using the BFI-K, which consists of nine questions. Fatigue and its interference with daily living were scored by patients on a scale from 0 to 10, which ranged from ‘no fatigue’ to ‘fatigue as bad as you can imagine.’ The internal consistency of the Korean version was very high (Cronbach's α coefficient=0.96), and validity was confirmed by factor analysis.¹⁶⁾ The BFI-K was evaluated for fatigue ‘now,’ ‘usual’ fatigue, and ‘worst’ fatigue during the last 24 hours. Aspects evaluated included general activity, mood, walking ability, normal work (both outside of the home and daily chores), relationships with other people, and enjoyment of life.¹⁷⁾ Since BFI-K scores for ‘usual’ fatigue are 4.07±2.27 in the general Korean population,¹⁾ only participants with a BFI-K score greater than four were included in the study. We assessed depression and sleep quality, using a self-evaluation questionnaire. The Beck Depression Inventory (BDI) is one of the most widely used self-reporting instruments for measuring the severity of depression. Each symptom is rated on a 4-point intensity scale, and scores are added to produce a total within the range 0 to 63. The BDI is useful in screening depression in the general population, as well as measuring the severity of depression among psychiatric patients.¹⁸⁾ The Pittsburgh Sleep Quality Index (PSQI) is used worldwide to evaluate sleep quality during the previous month. The Korean version of the PSQI has previously been tested for reliability (Cronbach's α coefficient=0.84), sensitivity (0.943), and specificity (0.844).¹⁹⁾ Depression was defined as a total BDI score of more than 13 points, while insomnia was defined as more than five points on the PSQI.

Buccal epithelial cells of all participants were collected from saliva, after rubbing oral mucosa against teeth. Saliva samples were washed by centrifuging at 1,200× g for 3 minutes in phosphate-buffered saline. DNA components were extracted using the Tissue Spin Mini Kit (Nucleogen, Ansan, Korea). The quantity and quality of DNA were measured with an Epoch spectrophotometer (BioTek, Winooski, VT, USA). The mtDNA content was measured by a real-time quantitative polymerase chain reaction (qPCR) for mitochondrial encoded nicotinamide adenine dinucleotide dehydrogenase 1 (MT-ND1), which was normalized to hemoglobin beta (HBB), encoded by nuclear DNA. Using the CFX96 Real-time System (Bio-Rad, Singapore, Singapore), qPCR was performed. The 15-µL reaction mixture contained 10 ng of template DNA, 1X Thunderbird SYBR qPCR Mix (Toyobo, Osaka, Japan) and 7.5 pmol of primers. Thermal cycling conditions consisted of 1 cycle at 95℃ for 60 seconds, followed by 35 cycles of: 95℃ for 5 seconds, 61℃ for 10 seconds, and 72oC for 20 seconds. After PCR, a dissociation curve analysis was performed from 61℃ to 95℃ (in 0.5℃ increments) to confirm specific amplification. To assess the efficiency of all primer pairs, the standard curves were derived from seven serial dilutions of reference DNA at 40, 20, 10, 5, 2.5, 1.25, and 0.625 ng, using the Bio-Rad CFX Manager version 1.6 software (Bio-Rad). Primers for MT-ND1 were 5'-GACCCTACTTCTAACCTCCCTGT-3' and 5'-TAGGAGGTGTATGAGTTGGTCGT-3'. HBB primers were 5'-ACCCAAGAGTCTTCTCTGTCTCCA-3' and 5'-TCTGCCGTTACTGCCCTGTG-3'. We also screened the University of California, Santa Cruz database (http://genome.ucsc.edu/) to confirm that the primer sequences were unique without any repeats. The threshold cycle number (Ct) indicates the fractional Ct, which is automatically determined for each primer. The ΔCt represented the relative abundance, calculated as {Ct (ND1)-Ct (HBB)}. The mtDNA was quantified using the equation: 2×2^−ΔCt.20) All experiments contained reference DNA and were repeated at least three times. Each experiment was normalized against serial dilutions of a control DNA sample.²¹⁾

To evaluate the relationship between mtDNA copy number and the characteristics measured using BFI-K, BDI, and PSQI, we compared these variables according to the tertile distribution of the log-transformed mtDNA copy number. All analyses used a logarithmically transformed mtDNA copy number.

Clinical characteristics, BFI-K, BDI, PSQI, and mtDNA copy number for the three groups (1) without combined symptoms, (2) with either insomnia or depression, and (3) with both insomnia and depression of fatigue patients were compared by Kruskal-Wallis or chi-squared test.

To assess the association between mtDNA copy number and fatigue according to combined symptoms, the percent difference (95% confidence interval) of mtDNA copy number per one point increase in BFI-K, BDI, or PSQI was calculated using a general linear model. The percent difference of mtDNA copy number was calculated by multiplying the value of (exponentiated β coefficient–1) by 100. Age, sex, BMI, smoking status, alcohol intake, physical exercise, and comorbidities were adjusted.

All analyses were performed using IBM PASW Statistics ver. 21.0 software (IBM Corp., Armonk, NY, USA), and a P-value less than 0.05 was considered statistically significant.