Noninvasive Type 2 Diabetes Screening: Clinical Evaluation of SCOUT DS in an Asian Indian Cohort

Abstract

Objective:

This study evaluated the noninvasive, point-of-care diabetes screening device, Scout DS^® (VeraLight Inc., Albuquerque, NM) (SCOUT), in a native Asian Indian cohort.

Research Design and Methods:

SCOUT is a tabletop, skin fluorescence spectrometer that reports a risk score following a 3–4-min noninvasive measurement of a subject's left volar forearm. SCOUT, fasting plasma glucose (FPG), and hemoglobin A_1c (A1C) were compared for detection of abnormal glucose tolerance (AGT) in a cohort of 256 subjects without previous diagnosis of diabetes or impaired glucose tolerance in Chennai, India. After an overnight fast, a 75-g, 2-h oral glucose tolerance test was administered, and AGT was defined as a plasma glucose value ≥140 mg/dL (7.8 mmol/dL). Sensitivity, false-positive rate (FPR), and receiver-operating characteristics area under the curve for AGT detection were computed for SCOUT, FPG, and A1C. Intra-day reproducibility of SCOUT was assessed.

Results:

SCOUT, FPG, and A1C (at respective thresholds of 50, 110 mg/dL, and 5.7%) exhibited sensitivities of 87%, 32%, and 86%, respectively, and FPR of 52%, 3%, and 58%, respectively. For the 177 subjects receiving a valid SCOUT Diabetes Score on both measurement attempts, the coefficient of variation was 5.8%, and the Pearson correlation was 0.91. A SCOUT score could be obtained on 91% of subjects after two attempts.

Conclusions:

The performance of SCOUT is similar to that of A1C, whereas FPG had a much lower sensitivity. SCOUT is an effective tool for AGT screening in Asian Indians.

Introduction

I n 2011, the International Diabetes Federation estimated that 61.3 million people in India 20–79 years old had type 2 diabetes mellitus, of whom 31.3 million were undiagnosed.¹ Also, 20.5 million had impaired glucose tolerance (IGT), a significant risk factor for future type 2 diabetes and cardiovascular disease. By 2030, without significant diabetes prevention efforts, the IDF estimates India will have 101.2 million people 20–79 years old with diabetes—a 65% increase—and another 32.2 million with IGT—a 57% increase.¹ A recent national study showed that the prevalence was also rapidly increasing in rural areas where facilities for screening are often unavailable.²

Asian Indian type 2 diabetes patients tend to develop the disease years earlier than western counterparts and also progress rapidly from IGT to diabetes.^3,4 Asian Indians with type 2 diabetes mellitus are more likely to die from heart disease but have lower rates of retinopathy, neuropathy, and nephropathy.³ Asian Indians are believed to be more genetically predisposed to type 2 diabetes, as they develop the disease at lower body mass indices than Europeans.^5,6

The Indian Diabetes Prevention Program demonstrated that lifestyle modification could reduce conversion from IGT to type 2 diabetes by 28.5% over approximately 3 years.⁷ Furthermore, the Indian Diabetes Prevention Program showed that pharmacological therapy with metformin reduced conversion to type 2 diabetes by 26.4% over 3 years.⁷

Screening at-risk individuals is therefore critical to primary diabetes prevention. Traditional blood-based measurements of glycemia are commonly used to screen for type 2 diabetes mellitus, but these methods have drawbacks that limit the percentage of the at-risk population that is tested. Barriers to effective diabetes screening programs using venous blood samples include blood-draw apprehension, requirements for overnight fasting, handling blood, disposing of sharps, lag time for laboratory processing of blood samples, and lack of qualified personnel and standardized laboratories, which make results unreliable.⁸ In India, random capillary glucose is frequently used as a screening tool. Although convenient, random capillary glucose suffers from poor accuracy in detecting prediabetes and either misses many cases or results in more false-positives.⁹ Because random capillary glucose requires blood, it creates barriers to testing (as noted above) and generates biohazards that could transmit diseases, such as hepatitis and human immunodeficiency virus, although the risk may be small.

Recently, noninvasive diabetes screening based on skin fluorescence has been proposed as an alternative to blood-based screening.^10
–12 The Scout DS^® (VeraLight Inc., Albuquerque, NM) (SCOUT) is a noninvasive diabetes screening device that uses fluorescence spectroscopy to measure diabetes biomarkers in the skin, including fluorescent advanced glycation end-products, such as pentosidine, and collagen cross-links as well as fluorescent intermediaries of cell metabolism and oxidative stress, such as NADH and FAD.^13,14 The tabletop device (Figure 1) illuminates the left volar forearm skin with low-intensity light at multiple near-ultraviolet and visible wavelengths. A specially designed fiber optic probe couples the excitation light to the underside of the subject's forearm near the elbow and relays resulting skin fluorescence and reflectance to a spectrograph and camera. SCOUT uses measured reflectance to assess light absorption due to melanin and hemoglobin in a subject's skin and incorporates this information into the measurement algorithm when analyzing skin fluorescence to produce a diabetes risk score.

FIG. 1.

Photographs of the SCOUT DS skin fluorescence spectrometer. The fluorescence of the left volar forearm skin is noninvasively measured by the fiber optic array visible in the top left.

To determine if diagnostic testing is necessary, the SCOUT can be used as a first step for noninvasive screening of individuals at risk for prediabetes and/or type 2 diabetes. Subjects need not fast before SCOUT measurement, no biohazards are generated, and point-of-service results are immediately available. Each valid subject measurement results in a SCOUT Diabetes Score (SDS) on a scale of 0–100. Higher SDS values suggest greater likelihood of type 2 diabetes. SDS results of ≥50 are considered a positive screen for prediabetes or type 2 diabetes, and diagnostic blood testing (oral glucose tolerance test [OGTT] or hemoglobin A_1c [A1C]) could then follow.

To date, SCOUT has been tested in multicenter clinical trials of North American populations.¹⁵ However, its applicability in tropical climates and Asian Indian populations is unknown. The Assessment of SCOUT in Chennai (ASC) trial is an investigation and prospective validation of SCOUT diabetes screening performance in an Asian Indian population that is at risk for type 2 diabetes and is of significance given the huge burden of diabetes and prediabetes in this ethnic group.

Subjects and Methods

A1C and fasting plasma glucose (FPG) were used as comparative screening methods. The “gold standard” was abnormal glucose tolerance (AGT), defined as a post-challenge plasma glucose of at least 140 mg/dL after a 75-g, 2-h OGTT.

The Madras Diabetes Research Foundation in Chennai, India, recruited 279 subjects from database of Dr. Mohan's Diabetes Specialties Centre, screening from the community and referrals from study participants after written informed consent was obtained from the subjects. The Institutional Ethics Committee of the Madras Diabetes Research Foundation approved the study protocol. Each subject visited the clinic once, after an overnight fast (water only for 8 h). The inclusion criterion was age ≥30 years of either sex. The exclusion criteria were as follows: known to have diabetes or prediabetes; receiving investigational treatments in the past 14 days; psychosocial issues that interfere with the ability to follow study procedures; conditions that cause secondary diabetes (Cushing's syndrome, acromegaly, hemochromatosis, pancreatitis, or cystic fibrosis); diagnosed with type 1 diabetes; pregnant women; receiving dialysis or having known renal compromise; scars, tattoos, rashes, or other disruption/discoloration on the left volar forearm; recent (within the past month) or current oral steroid therapy or topical steroids applied to the left forearm; chemotherapy within the past 12 months; receiving medications that fluoresce, including fluoroquinolones, tetracyclines, hydroxycholoroquine, or quinidine; and known to have, or at risk for, photosensitivity reactions (sensitive to ultraviolet light).

All subjects answered a short health history and had physical measurements taken (height, weight, waist circumference, and blood pressure). The subject was measured twice on SCOUT, with each measurement consisting of two successive forearm placements on the SCOUT device. Venipuncture was performed to collect FPG, A1C, hemogram, lipid profile, and liver function specimens. The subject then consumed a 75-g oral glucose load (82.5 g of glucose monohydrate in 250 mL of water) over a 5-min interval. At 120±10 min after consumption, venipuncture was performed to collect the 2-h post-challenge plasma glucose specimen. Plasma glucose and A1C assays were performed using an automated analyzer (model AU 2700/480; Beckman Coulter, Brea, CA) and testing system analyzers (Variant™ II Turbo; Bio-Rad, Hercules, CA). The A1C values were Diabetes Control and Complications Trial-aligned, and the laboratory is certified by the College of American Pathologists and the National Accreditation Board for Testing and Calibration. Creatinine, lipid profile, and liver function assays were performed using the Beckman Coulter model AU 2700/480 automated analyzer. The intra- and inter-assay coefficients of variation (CVs) for the plasma glucose assays were 2.2% and 2.5%, respectively; the corresponding values for the A1C assay were 0.6% and 1.5%.

SCOUT data were acquired using a CE-marked SCOUT commercial device (software revision 1.2), which might not report an SDS on subjects with very dark skin, fluorescent contamination on the skin (one known source is turmeric, topically applied by some Indian women), or inconsistent spectra between the two arm insertions. Subjects not receiving a SCOUT score after at least two attempts were deemed screen failures and did not continue with blood work at the first visit. A follow-up phase recalled screen failures for SCOUT remeasurement and re-acquisition of blood work. This facilitated the performance assessment of a new SCOUT algorithm that is more robust to dark skin and skin contamination.

VeraLight provided scores from the new SCOUT algorithm, which is under development and not yet commercially available. The new algorithm was independently calibrated using a separate dataset collected exclusively from a U.S. cohort, and ASC data were prospectively scored. The algorithm may not report a score on some subjects, most often because of inconsistent contact with the optical sensor between the two arm insertions.

Point estimates on sensitivity and false-positive rate (FPR) for AGT detection were calculated for FPG, A1C, and SDS, along with receiver-operator characteristic curves, area under the curve, negative predictive value, and positive predictive value. The 95% confidence intervals on FPR and sensitivity were calculated using the Wilson interval. For FPG, both the World Health Organization and American Diabetes Association-recommended thresholds for diabetes screening were evaluated (110 mg/dL and 100 mg/dL, respectively). For A1C, the thresholds recommended by the American Diabetes Association¹⁶ and the International Expert Committee¹⁷ were evaluated (5.7% and 6.0%, respectively). To assess the effectiveness of each screening test at early detection (prediabetes), the FPR and sensitivity were calculated for each of three prediabetes truth standards: IGT, which is OGTT ≥140 mg/dL and <200 mg/dL; impaired fasting glucose, which is FPG ≥100 mg/dL and <126 mg/dL; and “impaired” A1C, which is A1C ≥5.7% and <6.5%.

SDS reproducibility was assessed via intra-individual CV from the Hoorn study (Hoorn CV),¹⁸ given by \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} \begin{align*}Hoorn \ CV = \frac {SD_{\rm dif}/\sqrt {2}} {M_{\rm mean}} \times 100\end{align*} \end{document}

where SD _dif is the SD of the difference between each subject's first and second measurements and M _mean is the median of the mean of each subject's first and second measurements.

The relative skin tone of this cohort was assessed by summing the skin reflectance over the entire excitation and emission regions (360–660 nm) as assessed from the white light-emitting diode spectra acquired by SCOUT as part of the standard measurement. For comparison, the skin tone is plotted as a histogram distribution overlaid on that from the cohort of an at-risk screening cohort in the United States, which consisted of 59% white, 12% African American, and 27% Hispanic subjects and represents the full range of human skin tone.¹⁵

Statistical analyses were performed with MATLAB version 7.5 (R2007b, including the MATLAB statistics toolbox; The MathWorks, Natick, MA).

Results

Of the 279 enrolled subjects, 23 were excluded for the following reasons: consent error (n=5), problem ingesting glucose challenge (n=1), blood not collected at first visit and unable to return for follow-up (n=14), dropped out prior to blood draw (n=2), and did not attempt two SCOUT measurements (n=1). Of the remaining 256 evaluable subjects, 153 (60%) were normal, and 103 (40%) had AGT. As shown in Table 1, demographic variables with no statistically significant difference (per Pearson's χ² test or Wilcoxon rank sum test at 0.05 significance) between means of normal and AGT cohorts included gender, height, weight, body mass index, waist size, hemoglobin, low-density lipoprotein cholesterol, high-density lipoprotein cholesterol, and creatinine. Variables with significant difference (normal vs. AGT) included age, systolic blood pressure, diastolic blood pressure, and subject with a relative having diabetes.

Table 1.

Demographic Comparison of Normal and Abnormal Glucose Tolerance Cohorts

Variable	All (n=256)	NGT (n=153 [60%])	AGT (n=103 [40%])	P value
Age (years)	50.9±10.9	49.2±10.9	53.4±10.5	0.003^a
Female	130 (51%)	79 (52%)	51 (50%)	0.739^b
Blood pressure (mm Hg)
Systolic	132.7±18.6	129.1±17.7	138.0±18.7	<0.001^a
Diastolic	82.4±10.7	81.1±10.2	84.3±11.3	0.010^a
Hypertension	100 (39%)	44 (29%)	56 (54%)	<0.001^b
Parent with diabetes	101 (39%)	57 (37%)	44 (43%)	0.380^b
Sibling with diabetes	60 (23%)	31 (20%)	29 (28%)	0.144^b
Relative with diabetes	125 (49%)	67 (44%)	58 (56%)	0.049^b
Weight (kg)	66.7±11.8	66.3±12.2	67.2±11.3	0.690^a
Height (cm)	159.1±10.0	159.2±9.9	158.8±10.2	0.763^a
BMI (kg/m²)	26.4±4.2	26.2±4.3	26.7±3.9	0.387^a
Waist size (cm)	90.2±9.3	89.5±9.7	91.2±8.6	0.287^a
Waist size (cm)
Male	92.7±9.5	91.6±9.8	94.3±8.9	0.209^a
Female	87.8±8.5	87.5±9.3	88.2±7.2	0.927^a
SCOUT (SDS)	54.1±9.7	50.9±8.4	58.9±9.5	<0.001^a
SCOUT not measurable	22 (9%)	11 (7%)	11 (11%)	0.329^a
FPG (mg/dL)	100.8±26.7	91.9±8.1	113.9±37.3	<0.001^a
A1C (%)	6.1±1.1	5.7±0.4	6.8±1.5	<0.001^a
Hemoglobin (g/dL)	13.3±1.7	13.2±1.7	13.6±1.8	0.100^a
LDL (mg/dL)	116.5±35.0	117.5±35.7	115.0±34.0	0.854^a
HDL (mg/dL)	41.9±8.8	42.7±8.9	40.8±8.5	0.076^a
Total bilirubin (mg/dL)	0.7±0.3	0.7±0.3	0.7±0.3	0.261^a
Creatinine (mg/dL)	0.8±0.2	0.8±0.2	0.8±0.2	0.966^a
Red blood cell count (10⁶/μL)	4.7±0.5	4.7±0.5	4.7±0.5	0.560^a
White blood cell count (count/μL)	7202±1717	6982±1640	7529±1784	0.009^a
γ-Glutamyl transpeptidase (log₁₀) (IU/L)	1.3±0.2	1.3±0.2	1.4±0.2	0.002^a
Albumin (g/dL)	4.2±0.2	4.2±0.2	4.2±0.2	0.698^a

Data are expressed as either number (%) or mean±SD values. Abnormal glucose tolerance (AGT) and normal glucose tolerance (NGT) groups were stratified on the basis of an oral glucose tolerance test threshold of 140 mg/dL.

Wilcoxon rank sum test.

Pearson's χ² test.

A1C, hemoglobin A_1c; BMI, body mass index; FPG, fasting plasma glucose; HDL, high-density lipoprotein; LDL, low-density lipoprotein; SDS, SCOUT Diabetes Score.

Table 2 and Figure 2 show performance for each screening test at the respective thresholds (World Health Organization, American Diabetes Association, and VeraLight recommendations for FPG, A1C, and SCOUT, respectively). The sensitivities for SCOUT, A1C (at 5.7%), and FPG (at 110 mg/dL) were 87%, 86%, and 32%, respectively, while the FPRs were 52%, 58%, and 3%, respectively. The confidence intervals on both FPR and sensitivity for SCOUT versus A1C overlap, indicating no significant difference in performance in this cohort. SCOUT and A1C confidence intervals do not overlap with those of FPG, indicating superior sensitivity but inferior FPR for SCOUT and A1C relative to FPG. Although the area under the curve of SCOUT (75%) is slightly lower than for A1C (81%) or FPG (81%), the receiver-operator characteristic curves indicate that for FPR 40% and above, the receiver-operator characteristic performance of the three tests is very similar, and at the FPRs corresponding to the SCOUT and HbA1c thresholds, the three tests have nearly identical sensitivity. Table 3 shows that SCOUT maintains a high sensitivity and consistent FPR for the three alternate truth definitions for detection of prediabetes.

FIG. 2.

Receiver-operator characteristic (ROC) curve performance of the screening tests. AGT, abnormal glucose tolerance; FPG, fasting plasma glucose; HbA1c, hemoglobin A_1c; SDS, SCOUT Diabetes Score.

Table 2.

Screening Test Performance for Detection of Abnormal Glucose Tolerance, Including 95% Confidence Intervals, on False-Positive Rate and Sensitivity

Test	Threshold	AUC	FPR	Sensitivity	PPV	NPV
FPG (WHO)	110	0.81	0.03 (0.01–0.07)	0.32 (0.23–0.42)	0.88	0.69
FPG (ADA)	100	0.81	0.15 (0.10–0.22)	0.63 (0.53–0.72)	0.72	0.78
A1C	5.7	0.81	0.58 (0.50–0.66)	0.86 (0.77–0.92)	0.49	0.82
A1C	6.0	0.81	0.25 (0.18–0.32)	0.72 (0.62–0.80)	0.65	0.80
SCOUT	50	0.75	0.52 (0.44–0.60)	0.87 (0.79–0.92)	0.52	0.85

Abnormal glucose tolerance was defined by an oral glucose tolerance test of ≥140 mg/dL.

A1C, hemoglobin A_1c; ADA, American Diabetes Association; AUC, area under the curve; FPG, fasting plasma glucose; FPR, false-positive rate; NPV, negative predictive value; PPV, positive predictive value; WHO, World Health Organization.

Table 3.

Point-Estimate Performance of the Three Screening Tests for Detection of Prediabetes as Defined by Three Alternate Definitions

		IGT		IFG		Impaired A1C
Test	Threshold	FPR	Sensitivity	FPR	Sensitivity	FPR	Sensitivity
FPG	110 mg/dL	0.03	0.08			0.03	0.04
FPG	100 mg/dL	0.15	0.44			0.14	0.28
A1C	5.7%	0.58	0.80	0.59	0.84
A1C	6%	0.25	0.58	0.26	0.70
SCOUT	50	0.52	0.90	0.58	0.81	0.55	0.66

The three alternate definitions were impaired glucose tolerance (IGT) with oral glucose tolerance test of ≥140 mg/dL and <200 mg/dL, impaired fasting glucose (IFG) with fasting plasma glucose (FPG) of ≥100 mg/dL and <126 mg/dL, and impaired hemoglobin A_1c (A1C) of ≥5.7% and <6.5%.

FPR, false-positive rate.

The skin tone of this cohort is relatively dark, as demonstrated by Figure 3. The distribution of skin reflectance in this cohort overlaps with the darkest third of the diverse at-risk cohort from the United States shown for reference.

FIG. 3.

Distribution of skin reflectance (i.e., skin tone) in this cohort (Assessment of SCOUT in Chennai [ASC]) relative to a sample at-risk cohort in the United States (US) consisting of white, Hispanic, and African American subjects. Skin reflectance is dominated by melanin concentration and is therefore a measure of skin tone. This cohort consisted of dark- to very dark-skinned individuals. AU, arbitrary units.

For the 177 subjects receiving a valid SDS on both measurement attempts, the Hoorn CV was 5.8%, and the Pearson correlation was 0.91 with a value of P<0.001. Of the subjects examined, 79.3% received an SDS on the first attempt, and 91.4% received an SDS after two attempts. The performance metrics for SCOUT use the first SDS for each subject in two attempts. Subjects not receiving a SDS after two attempts were excluded from performance assessment on all three tests. The mean measurement time per subject (arm on instrument) was 217 s.

Discussion

This cohort is the first study of SCOUT screening performance on an Asian Indian cohort—a very challenging cohort for skin fluorescence technologies because of generally dark skin tone. Despite this, the SCOUT generated a score on 91% of subjects with no more than two measurement attempts per subject and achieved AGT screening performance on par with HbA1c in this cohort. The primary reason that SCOUT does not report a score on a subject is spectral disagreement between the two arm measurements, which can occur because of several reasons, including inconsistent arm placement on the device or heterogeneous optical skin properties (i.e., wrinkles, freckles, or uneven melanin).

This study identified topical use of turmeric as an interferent with the SCOUT (or any other fluorescence) measurement. The device did not report measurements on three subjects who reported recent or habitual turmeric use on skin at the SCOUT measurement site.

The SCOUT results shown in this work were generated from the skin spectra using a new proprietary SCOUT algorithm currently under development by VeraLight, the manufacturer of SCOUT. The new algorithm was independently developed on a dataset consisting exclusively of subjects in the United States; although this dataset did not include any Asian Indian subjects, it did have a large fraction of American dark-skinned subjects of similar skin tone to the Chennai cohort of this study.

In this study, we report that the sensitivity of SCOUT, A1C, and FPG to detect abnormal glucose tolerance was 87%, 86%, and 32%, respectively. In a recent study done in the United States, SCOUT performed with 68% sensitivity at a 38% FPR and had equivalent sensitivity to FPG and A1C in the 20–50% FPR range.¹⁵ In this context the finding in Chennai that SCOUT has a 87% sensitivity to detect AGT is an important observation because of earlier concerns about the accuracy of fluorescent measurements in darkly pigmented skin, which posed questions about its widespread use as a diabetes screening tool. However, it is now confirmed that the instrument compensates for skin pigmentation so that performance is not diminished by skin coloration as evident from the present study in Chennai, which demonstrated a SCOUT score in almost 91% of subjects.

Although the 52% false positives for SCOUT generates a large number of false positives, this is not necessarily undesirable for population screening. Chatterjee et al.¹⁹ found that even screening tests with false positives as high as 43% were economically superior to the case of no or limited screening, after accounting for the costs of screening and therapeutic intervention. This is because the cost of a false negative (delayed treatment of diabetes and associated complications) is significantly higher than that of a false positive (confirmatory testing). An economic analysis of widespread screening in India would help assess whether the SCOUT threshold of 50 is economically desirable in India.

In our study, the sensitivity of FPG to detect glucose intolerance is only 32%. Earlier studies of FPG screening for undiagnosed diabetes have also reported that sensitivities range only from 40% to 65%.²⁰ The large false-negative rate for FPG testing is a latent problem and contributes to the growing number of undiagnosed cases of prediabetes and type 2 diabetes. It is interesting that, in our study, the sensitivity of SCOUT and A1C to detect AGT appear similar. Using the A1C cutpoint, Mohan et al.²¹ have demonstrated that the sensitivity of the A1C test to detect type 2 diabetes was >88%. However, the sensitivity of A1C to detect impaired fasting glucose and IGT was only 60% and 65.6%, respectively. These studies emphasize the role of SCOUT as a promising point-of-care tool to detect subjects with risk for type 2 diabetes.

There are many sources of variance that can affect the optical properties of skin, and thus the spectra acquired in a skin fluorescence measurement. The major sources of optical variation include absorption from melanin and hemoglobin and the scattering properties determined by collagen/elastin organization and the optical interface with the device. Other than the fasting requirement on subjects due to the simultaneous acquisition of an OGTT (and associated morning timing of the visit), the only environmental or physiological factor controlled in this study was the ambient temperature in the room with the SCOUT instrument, which was climate-controlled with an air conditioner to be between 25°C and 30°C. Therefore, any modifications of skin optical properties due to physiological or environmental effects are inherent in the performance results.

In conclusion, SCOUT has several advantages over FPG and/or A1C for mass diabetes screening: SCOUT does not require a blood draw, does not require fasting, can be done at any time of day, is simple to operate, uses no reagents or disposables (other than cleaning supplies), and provides point-of-care results in under 4 min for most subjects. In addition, the cost per test for SCOUT in mass population screening is competitive with that of the blood tests. SCOUT therefore eliminates some obstacles to mass population screening for diabetes, while potentially offering accuracy comparable with A1C for screening for AGT in South Asia, which is currently the epicenter of the global diabetes epidemic.¹

Footnotes

Acknowledgments

We are grateful to Mr. Anand Kumar, Ms. Neeraja, Ms. Ramya, Ms. Gini Venisha, Ms. Lavanya, Mr. Jagan Kumar, and Ms. Baby from the epidemiology team at the Madras Diabetes Research Foundation for the fieldwork and, most importantly, the subjects who participated in the study. We thank Dr. Ranjani Harish and Ms. N. Lakshmi of Translational Research Department of the Madras Diabetes Research Foundation for their help. We are grateful to the VeraLight, Inc. for the financial support provided for the study. Financial assistance from the Department of Biotechnology, New Delhi, is also gratefully acknowledged.

Author Disclosure Statement

N.I.M. is an employee and stock option holder of VeraLight, Inc. V.M., C.S.S.R., B.S.R., B.M., R.A., S.P., M.D., and R.P. declare no competing financial interests exist.

References

Unwin

, Whiting

, Guariguata

, Ghyoot

, Gan

. Diabetes Atlas, 5 ^th. Brussels: International Diabetes Federation, 2011.

Anjana

, Pradeepa

, Deepa

, Datta

, Sudha

, Unnikrishnan

, Bhansali

, Joshi

, Yajnik

, Dhandhania

, Nath

, Das

, Rao

, Madhu

, Shukla

, Kaur

, Priya

, Nirmal

, Parvathi

, Subhashini

, Subashini

, Ali

, Mohan

. ICMR–INDIAB Collaborative Study Group: Prevalence of diabetes and prediabetes (impaired fasting glucose and/or impaired glucose tolerance) in urban and rural India: Phase I results of the Indian Council of Medical Research–INdia DIABetes (ICMR–INDIAB) study. Diabetologia, 2011; 54:3022–3027.

Mohan

, Sandeep

, Deepa

, Shah

, Varghese

. Epidemiology of type 2 diabetes; Indian scenario. Indian J Med Res, 2007; 125:217–230.

Mohan

, Deepa

, Shanthirani

, Farooq

, Ganesan

, Datta

. Secular trends in the prevalence of diabetes and impaired glucose tolerance in urban South India—the Chennai Urban Rural Epidemiology Study (CURES-17) Diabetologia, 2006; 49:1175–1178.

Huxley

, James

, Barzi

, Patel

, Lear

, Suriyawongpaisal

, Janus

, Caterson

, Zimmet

, Prabhakaran

, Reddy

, Woodward

. Obesity in Asia Collaboration. Ethnic comparisons of the cross-sectional relationships between measures of body size with diabetes and hypertension. Obes Rev, 2008; 9,Suppl 1:53–61.

Lear

, Humphries

, Kohli

, Chockalingam

, Frohlich

, Birmingham

. Visceral adipose tissue accumulation differs according to ethnic background: results of the Multicultural Community Health Assessment Trial (M-CHAT) Am J Clin Nutr, 2007; 86:353–359.

Ramachandran

, Snehalatha

, Mary

, Mukesh

, Bhaskar

, Vijay

. The Indian Diabetes Prevention Programme shows that lifestyle modification and metformin prevent type 2 diabetes in Asian Indian subjects with impaired glucose tolerance (IDPP-1) Diabetologia, 2006; 49:289–297.

Anjana

, Pradeepa

, Deepa

, Datta

, Sudha

, Unnikrishnan

, Nath

, Das

, Madhu

, Rao

, Shukla

, Kaur

, Ali

, Mohan

. The Indian Council of Medical Research–India Diabetes (ICMR–INDIAB) Study: methodological details. J Diabetes Sci Technol, 2011; 5:906–914.

Somannavar

, Ganesan

, Deepa

, Datta

, Mohan

. Random capillary blood glucose cut points for diabetes and prediabetes derived from community-based opportunistic screening in India. Diabetes Care, 2010; 32:641–643.

10.

Ediger

, Olson

, Maynard

. Noninvasive optical screening for diabetes. J Diabetes Sci Technol, 2009; 3:776–780.

11.

Hull

, Ediger

, Unione

, Deemer

, Stroman

, Baynes

. Noninvasive, optical detection of diabetes: model studies with porcine skin. Opt Express, 2004; 12:4496–4510.

12.

Bos

, de Ranitz-Greven

, de Valk

. Advanced glycation end products, measured as skin autofluorescence and diabetes complications: a systematic review. Diabetes Technol Ther, 2011; 13:773–779.

13.

Yoshino

, Mills

, Yoon

, Imai

. Nicotinamidemononucleotide, a key NAD⁺ intermediate, treats the pathophysiology of diet- and age-induced diabetes in mice. Cell Metab, 2011; 14:528–536.

14.

Heikal

. Intracellular coenzymes as natural biomarkers for metabolic activities and mitochondrial anomalies. Biomark Med, 2010; 4:241–263.

15.

Olson

, Matter

, Ediger

, Way

, Hull

, Maynard

. Clinical validation of the SCOUT DS^® noninvasive prediabetes and diabetes screening device [abstract] Diabetes, 2012; 61:A348.

16.

American Diabetes Association. Standards of medical care in diabetes—2010. Diabetes Care, 2010; 33,Suppl 1:S11–S61.

17.

International Expert Committee. International Expert Committee report on the role of the A1C assay in the diagnosis of diabetes. Diabetes Care, 2009; 32:1327–1334.

18.

Mooy

, Grootenhuis

, de Vries

, Kostense

, Popp-Snijders

, Bouter

, Heine

. Intra-individual variation of glucose, specific insulin and proinsulin concentrations measured by two oral glucose tolerance tests in a general Caucasian population: the Hoorn Study. Diabetologia, 1996; 39:298–305.

19.

Chatterjee

, Lipscomb

, Narayan

VKM

, Phillips

. Screening adults for pre-diabetes and diabetes may be cost-saving. Diabetes Care, 2010; 33:1484–1490.

20.

Engelgau

, Narayan

, Herman

. Screening for type 2 diabetes. Diabetes Care, 2000; 23:1563–1580.

21.

Mohan

, Vijayachandrika

, Gokulakrishnan

, Anjana

, Ganesan

, Weber

, Venkat Narayan

. HbA1c cut points to define various glucose intolerance groups in Asian Indians. Diabetes Care, 2010; 33:515–519.