Corneal confocal microscopy for identification of diabetic sensorimotor polyneuropathy: a pooled multinational consortium study (original) (raw)

• Journal List
• Springer
• PMC6061173

Diabetologia. 2018; 61(8): 1856–1861.

Bruce A. Perkins,1,2 Leif E. Lovblom,1 Vera Bril,3 Daniel Scarr,1 Ilia Ostrovski,1 Andrej Orszag,1 Katie Edwards,4 Nicola Pritchard,4 Anthony Russell,5 Cirous Dehghani,4 Danièle Pacaud,6 Kenneth Romanchuk,7 Jean K. Mah,6 Maria Jeziorska,8 Andrew Marshall,9 Roni M. Shtein,10 Rodica Pop-Busui,11 Stephen I. Lentz,11 Andrew J. M. Boulton,12 Mitra Tavakoli,9,13 Nathan Efron,4 and Rayaz A. Malik9,14

Bruce A. Perkins

1Lunenfeld-Tanenbaum Research Institute, Leadership Sinai Centre for Diabetes, Mount Sinai Hospital, L5209-60 Murray Street, Box 16, Toronto, ON M5T 3L9 Canada

2Division of Endocrinology and Metabolism, Department of Medicine, University of Toronto, Toronto, ON Canada

Leif E. Lovblom

1Lunenfeld-Tanenbaum Research Institute, Leadership Sinai Centre for Diabetes, Mount Sinai Hospital, L5209-60 Murray Street, Box 16, Toronto, ON M5T 3L9 Canada

Vera Bril

3The Ellen and Martin Prosserman Centre for Neuromuscular Diseases, Krembil Neuroscience Centre, Division of Neurology, Department of Medicine, University Health Network, University of Toronto, Toronto, ON Canada

Daniel Scarr

1Lunenfeld-Tanenbaum Research Institute, Leadership Sinai Centre for Diabetes, Mount Sinai Hospital, L5209-60 Murray Street, Box 16, Toronto, ON M5T 3L9 Canada

Ilia Ostrovski

1Lunenfeld-Tanenbaum Research Institute, Leadership Sinai Centre for Diabetes, Mount Sinai Hospital, L5209-60 Murray Street, Box 16, Toronto, ON M5T 3L9 Canada

Andrej Orszag

1Lunenfeld-Tanenbaum Research Institute, Leadership Sinai Centre for Diabetes, Mount Sinai Hospital, L5209-60 Murray Street, Box 16, Toronto, ON M5T 3L9 Canada

Katie Edwards

4Institute of Health and Biomedical Innovation, Queensland University of Technology, Brisbane, QLD Australia

Nicola Pritchard

4Institute of Health and Biomedical Innovation, Queensland University of Technology, Brisbane, QLD Australia

Anthony Russell

5Faculty of Medicine, University of Queensland, Woolloongabba, QLD Australia

Cirous Dehghani

4Institute of Health and Biomedical Innovation, Queensland University of Technology, Brisbane, QLD Australia

Danièle Pacaud

6Alberta Children’s Hospital, Department of Paediatrics, Cumming School of Medicine, University of Calgary, Calgary, AB Canada

Kenneth Romanchuk

7Department of Surgery, University of Calgary, Calgary, AB Canada

Jean K. Mah

6Alberta Children’s Hospital, Department of Paediatrics, Cumming School of Medicine, University of Calgary, Calgary, AB Canada

Maria Jeziorska

8Division of Cardiovascular Sciences, School of Medical Sciences, Faculty of Biology, Medicine and Health, University of Manchester, Manchester, UK

Andrew Marshall

9Centre for Endocrinology and Diabetes, School of Medical Sciences, Faculty of Biology, Medicine and Health, University of Manchester, Oxford Road, Manchester, M13 9PL UK

Roni M. Shtein

10Ophthalmology & Visual Science Department, University of Michigan, Ann Arbor, MI USA

Rodica Pop-Busui

11Department of Internal Medicine, Division of Metabolism, Endocrinology & Diabetes, University of Michigan, Ann Arbor, MI USA

Stephen I. Lentz

11Department of Internal Medicine, Division of Metabolism, Endocrinology & Diabetes, University of Michigan, Ann Arbor, MI USA

Andrew J. M. Boulton

12Division of Diabetes, Endocrinology & Gastroenterology, School of Medical Sciences, Faculty of Biology, Medicine and Health, University of Manchester, Manchester, UK

Mitra Tavakoli

9Centre for Endocrinology and Diabetes, School of Medical Sciences, Faculty of Biology, Medicine and Health, University of Manchester, Oxford Road, Manchester, M13 9PL UK

13Diabetes, Vascular Research Centre, Institute of Health Research, University of Exeter Medical School, Exeter, UK

Nathan Efron

4Institute of Health and Biomedical Innovation, Queensland University of Technology, Brisbane, QLD Australia

Rayaz A. Malik

9Centre for Endocrinology and Diabetes, School of Medical Sciences, Faculty of Biology, Medicine and Health, University of Manchester, Oxford Road, Manchester, M13 9PL UK

14Department of Medicine, Weill Cornell Medicine-Qatar, Doha, Qatar

1Lunenfeld-Tanenbaum Research Institute, Leadership Sinai Centre for Diabetes, Mount Sinai Hospital, L5209-60 Murray Street, Box 16, Toronto, ON M5T 3L9 Canada

2Division of Endocrinology and Metabolism, Department of Medicine, University of Toronto, Toronto, ON Canada

3The Ellen and Martin Prosserman Centre for Neuromuscular Diseases, Krembil Neuroscience Centre, Division of Neurology, Department of Medicine, University Health Network, University of Toronto, Toronto, ON Canada

4Institute of Health and Biomedical Innovation, Queensland University of Technology, Brisbane, QLD Australia

5Faculty of Medicine, University of Queensland, Woolloongabba, QLD Australia

6Alberta Children’s Hospital, Department of Paediatrics, Cumming School of Medicine, University of Calgary, Calgary, AB Canada

7Department of Surgery, University of Calgary, Calgary, AB Canada

8Division of Cardiovascular Sciences, School of Medical Sciences, Faculty of Biology, Medicine and Health, University of Manchester, Manchester, UK

9Centre for Endocrinology and Diabetes, School of Medical Sciences, Faculty of Biology, Medicine and Health, University of Manchester, Oxford Road, Manchester, M13 9PL UK

10Ophthalmology & Visual Science Department, University of Michigan, Ann Arbor, MI USA

11Department of Internal Medicine, Division of Metabolism, Endocrinology & Diabetes, University of Michigan, Ann Arbor, MI USA

12Division of Diabetes, Endocrinology & Gastroenterology, School of Medical Sciences, Faculty of Biology, Medicine and Health, University of Manchester, Manchester, UK

13Diabetes, Vascular Research Centre, Institute of Health Research, University of Exeter Medical School, Exeter, UK

14Department of Medicine, Weill Cornell Medicine-Qatar, Doha, Qatar

Corresponding author.

Received 2018 Jan 10; Accepted 2018 Apr 6.

Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.

Abstract

Aims/hypothesis

Small cohort studies raise the hypothesis that corneal nerve abnormalities (including corneal nerve fibre length [CNFL]) are valid non-invasive imaging endpoints for diabetic sensorimotor polyneuropathy (DSP). We aimed to establish concurrent validity and diagnostic thresholds in a large cohort of participants with and without DSP.

Methods

Nine hundred and ninety-eight participants from five centres (516 with type 1 diabetes and 482 with type 2 diabetes) underwent CNFL quantification and clinical and electrophysiological examination. AUC and diagnostic thresholds were derived and validated in randomly selected samples using receiver operating characteristic analysis. Sensitivity analyses included latent class models to address the issue of imperfect reference standard.

Results

Type 1 and type 2 diabetes subcohorts had mean age of 42 ± 19 and 62 ± 10 years, diabetes duration 21 ± 15 and 12 ± 9 years and DSP prevalence of 31% and 53%, respectively. Derivation AUC for CNFL was 0.77 in type 1 diabetes (p < 0.001) and 0.68 in type 2 diabetes (p < 0.001) and was approximately reproduced in validation sets. The optimal threshold for automated CNFL was 12.5 mm/mm2 in type 1 diabetes and 12.3 mm/mm2 in type 2 diabetes. In the total cohort, a lower threshold value below 8.6 mm/mm2 to rule in DSP and an upper value of 15.3 mm/mm2 to rule out DSP were associated with 88% specificity and 88% sensitivity.

Conclusions/interpretation

We established the diagnostic validity and common diagnostic thresholds for CNFL in type 1 and type 2 diabetes. Further research must determine to what extent CNFL can be deployed in clinical practice and in clinical trials assessing the efficacy of disease-modifying therapies for DSP.

Electronic supplementary material

The online version of this article (10.1007/s00125-018-4653-8) contains peer-reviewed but unedited supplementary material, which is available to authorised users.

Keywords: Corneal confocal microscopy, Corneal nerves, Diabetic neuropathy, Diabetic sensorimotor polyneuropathy, Small nerve fibre morphology

Introduction

Diabetic sensorimotor polyneuropathy (DSP) occurs in 50–90% of people with diabetes and is a progressive, length-dependent process of nerve injury with complex underlying causal mechanisms [1]. Because of the long subclinical latency period, early identification and management could potentially limit the morbidity and healthcare costs of advanced neuropathy with its associated pain, foot deformity, ulceration and amputation. The diagnosis of DSP is often made late, as neurological and electrophysiological testing of large myelinated fibres identifies established neuropathy [1, 2]. Early identification of unmyelinated small nerve fibre injury will likely provide the best opportunity for effective therapy [1, 3].

Small cohort studies have shown that in vivo corneal confocal microscopy (IVCCM) is an objective and reproducible means to quantify small fibre damage [3]. The rapid non-invasive nature of this procedure and automated image analysis may enable eye specialists to perform this procedure alongside routine examination for diabetic retinopathy [4–7]. However, small cohort studies can be biased in participant selection, in IVCCM image acquisition, in corneal nerve quantification and in defining DSP. We have undertaken a stratified cross-sectional multicentre pooled analysis of 998 participants with diabetes to more definitively establish the diagnostic validity of IVCCM for DSP using manual and automated analysis techniques.

Methods

Study population

Five hundred and sixteen people with type 1 diabetes mellitus (432 adults and 84 adolescents) and 482 adults with type 2 diabetes mellitus with and without DSP (total N = 998) were studied between 2008 and 2011. Participant-level data was pooled from five separate cohorts as part of a prospective study of diagnostic validity by an international consortium (National Institutes of Health [NIH] grant 1DP3DK104386-01, ClinicalTrials.gov registration no. NCT02423434). Two preliminary reports of diagnostic accuracy from individual centres have previously been published [5, 6]; 190 participants from these two studies are included in this current analysis, representing 19% of this 998-person study. Additional details are provided in electronic supplementary material (ESM) Methods.

Study design

This is a cross-sectional analysis of baseline visits, reported according to the 2015 Standards for Reporting of Diagnostic Accuracy statement [8]. The diagnostic index test was quantification of corneal nerve morphology obtained by IVCCM, the target condition was DSP, and the reference standard was based on the Toronto consensus criteria incorporating electrophysiological abnormality in the lower limbs [1]. The index test and reference standard were conducted during the same study visit; staff performing the reference standard were blinded to results of the index test (and vice versa). For the index test, participants underwent examination of the sub-basal nerve plexus of the cornea using the Heidelberg Tomograph Rostock Cornea Module III (Heidelberg Engineering GmbH, Heidelberg, Germany and Heidelberg Engineering, Smithfield, RI, USA) according to published methods [9]. Using a manual (MANUAL) and automated (AUTO) protocol [4], corneal nerve fibre length (CNFL), corneal nerve branch density (CNBD) and corneal nerve fibre density (CNFD) were quantified. Published data have demonstrated similar cohort IVCCM characteristics, reproducibility and validity, regardless of study centre. Full details of the index test and reference standard are provided in ESM Methods.

Statistics

Analysis was stratified by diabetes type and included derivation and validation sets. Baseline characteristics were compared using simple univariable statistics. Receiver operating characteristic (ROC) curves were generated and the AUC, representing diagnostic accuracy, was compared. Optimal diagnostic thresholds were identified by distance to the point of perfect discrimination. Simple random sampling, without replacement with an equal proportion of centre membership, was used to create derivation and validation sets. The following validation criteria were used: (1) validation AUC fell inside the 95% CI of the derivation AUC and (2) the optimal thresholds of the derivation set had similar characteristics to the validation set. The AUC of each test was also compared between subcohorts. An α level of 0.05 was used (two-tailed). Sensitivity analyses were undertaken to account for possible imperfect reference standard and included modification of the reference standard variables to create less- and more-stringent definitions, composite reference standard methods and latent class analysis. ROC regression was used to determine the effects of age and sex on diagnostic accuracy. Alternative diagnostic thresholds were investigated. Additional details of the statistics, and variables used for sensitivity analyses, are provided in ESM Methods.

Results

Among eligible participants enrolled at the five centres, 516/574 (90%) with type 1 diabetes and 482/527 (91%) with type 2 diabetes underwent the index test and reference standard (ESM Fig. 1). General characteristics of the study population and the diabetes subcohorts are shown in Table ​1. DSP was present in 415 (42%) of the study population, in 160 (31%) of the type 1 diabetes subcohort and in 255 (53%) of the type 2 diabetes subcohort. Significantly impaired nerve conduction and IVCCM variables were observed in participants with vs without DSP; the presence of a broad spectrum of neuropathy measures was confirmed (ESM Tables 1 and 2).

Table 1

Characteristics of the 998 study participants

Fig. ​1 displays the ROC curves for IVCCM quantified by the automated protocol in the type 1 diabetes (Fig. ​1a) and type 2 diabetes (Fig. ​1b) derivation sets. In type 1 diabetes, CNFLAUTO had an AUC of 0.77 and an optimal threshold of 12.5 mm/mm2 (73% sensitivity and 69% specificity). In type 2 diabetes, CNFLAUTO had an AUC of 0.68 and an optimal threshold of 12.3 mm/mm2 (69% sensitivity and 63% specificity). In both type 1 and type 2 diabetes derivation sets, AUC for CNFLAUTO was significantly greater than 0.50 (which represents a test with no diagnostic accuracy, p < 0.001 for both comparisons). CNFLAUTO was associated with the highest AUC among the IVCCM variables in both subcohorts (ESM Table 3). Similar results were observed for IVCCM variables quantified by the manual protocol, and results for all variables were generally confirmed in the validation sets. Full details of the ROC curve analysis are presented in ESM Table 3.

Determination of diagnostic accuracy and optimal thresholds for identification of DSP by IVCCM in the derivation sets. (a) Optimal threshold for CNFLAUTO in type 1 diabetes was 12.5 mm/mm2, 73% sensitivity and 69% specificity, positive predictive value 50%, negative predictive value 86%, positive likelihood ratio 2.32 and negative likelihood ratio 0.39. (b) Optimal threshold for CNFLAUTO in type 2 diabetes was 12.3 mm/mm2, 69% sensitivity and 63% specificity, positive predictive value 66%, negative predictive value 66%, positive likelihood ratio 1.86, and negative likelihood ratio 0.49. Continuous black lines, CNFLAUTO; grey lines, CNBDAUTO; dashed black lines, CNFDAUTO. AUC values for CNFLAUTO, CNBDAUTO and CNFDAUTO were 0.77, 0.73 and 0.71 in type 1 diabetes, respectively, and 0.68, 0.66 and 0.52 in type 2 diabetes, respectively. The p value for comparison of AUC for CNFLAUTO between type 1 and type 2 diabetes derivation sets was not significant at 0.060; when the derivation and validations sets were combined, this p value was 0.003. T1DM, type 1 diabetes; T2DM, type 2 diabetes

Although the AUC values differed marginally by diabetes type, the optimal thresholds were virtually identical. We thus determined diagnostic accuracy in the full 998-person study (ESM Table 3): CNFLAUTO had an AUC of 0.71 and an optimal threshold of 12.3 mm/mm2 (67% sensitivity, 66% specificity, 59% positive predictive value, 74% negative predictive value, 1.97 positive likelihood ratio and 0.50 negative likelihood ratio). CNFLMANUAL had marginally lower AUC (0.70, p = 0.006 vs CNFLAUTO) but its optimal threshold value of 16.3 mm/mm2 had similar operating characteristics. The alternative threshold analysis, in which upper and lower threshold values were used to simultaneously maximise sensitivity and specificity, is shown in ESM Table 4. We noted that in the 998-person group, a lower CNFLAUTO threshold value of <8.6 mm/mm2 to rule in DSP and an upper CNFLAUTO threshold value of 15.3 mm/mm2 to rule out DSP was associated with 88% specificity and 88% sensitivity.

The sensitivity analyses are summarised in ESM Fig. 2 and ESM Table 5. In type 1 diabetes, more-stringent reference standard definitions resulted in higher AUC for CNFL. Performance using the composite reference test and latent class analysis for DSP case definition resulted in higher AUC (though differences were not statistically significant). No differences were observed in type 2 diabetes. No statistically significant effects of age or sex on ROC curves were found for CNFL.

Discussion

The findings of this large multicentre pooled concurrent diagnostic validity study reveal that IVCCM had diagnostic validity despite an imperfect reference standard for DSP, using both manual and automated corneal nerve quantification; CNFL was the optimal IVCCM variable and the estimate of performance in the primary analysis was conservative compared with sensitivity analyses that addressed the issue of the imperfect reference standard.

An objective imaging biomarker that can identify early-stage DSP (when interventions are most likely to be effective) and that can be used as an appropriate endpoint in the evaluation of putative therapies does not currently exist [1]. Late diagnosis limits the potential benefits of early risk factor management in preventing neuropathy-related sequelae [2]. The diagnosis of DSP itself is controversial as no definitive gold-standard testing exists aside from electrophysiological evaluation, which primarily identifies later-stage, large-fibre dysfunction and requires considerable specialist expertise, resources and time. In this context, IVCCM represents a rapid, non-invasive imaging endpoint for identifying early small fibre neuropathy. It has been extensively studied in small cross-sectional and cohort studies, which have established normative distributions [10], feasibility, reproducibility and the impact of variations in equipment and procedures.

As electrophysiological testing identifies later-stage rather than early-stage neuropathy, in the current analysis subclinical levels of neuropathy that were not classified as cases might a priori be expected to accentuate false-positives and impair test specificity. We believe this is the major reason for not achieving conventional standards of diagnostic performance and operating characteristics in this study. However, the performance and thresholds are sufficient to raise confidence in automated IVCCM as a diagnostic test [4]. Further research must focus on evaluation of the influence of IVCCM on treatment decisions, possible roles relative to existing tests, its impact on clinical outcomes such as new onset symptomatic neuropathy and foot complications, its role in further evaluation of therapies for neuropathy and its economic impact.

The present study minimised common sources of bias in diagnostic studies, such as recruitment, spectrum and verification bias, but it had limitations. Though common protocols were used, centralised supervision of IVCCM image acquisition and analysis and electrophysiological testing were not implemented. As a cross-sectional analysis, it did not evaluate the predictive validity of IVCCM (a future goal of the consortium). Confirmation of a lack of age effect will require a larger older-adult sample size.

The diagnostic utility of IVCCM has been established in the largest cohort to date and the findings of this study further support the notion that IVCCM is an objective and simple diagnostic test for DSP. Further research must determine to what extent IVCCM can be deployed in clinical practice and in clinical trials assessing the efficacy of disease-modifying therapies for DSP.

Electronic supplementary material

Acknowledgements

We acknowledge the generous support of Randy and Jenny Frisch and The Harvey and Annice Frisch Family Fund and the contributions of the Steven and Ofra Menkes Fund, who supported work at the University of Toronto. Parts of this study were presented in abstract form at the following meetings: the American Diabetes Association 77th Scientific Sessions, San Diego, CA, USA, June 2017; the Canadian Diabetes Association and the Canadian Society of Endocrinology and Metabolism 19th Annual Professional Conference and Annual Meetings, Ottawa, ON, Canada, October 2016 and 26th Annual Meeting of the Diabetic Neuropathy Study Group of the EASD (NEURODIAB), Bucharest, Romania, September 2016.

Contribution statement

All authors participated in data acquisition and interpretation, reviewed the manuscript for scholarly content and accuracy and gave approval for the final draft. BAP, NE and RAM designed the study. BAP, LEL and RAM contributed to data interpretation and wrote the first draft of the manuscript. LEL was responsible for statistical analysis, with contributions from DS and IO. BAP had full access to all the data in the study and had final responsibility for the decision to submit for publication.

Abbreviations

Funding

This report was supported by the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) of the NIH (grant no. 1DP3DK104386-01). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. Initially, four of the study centres were funded by JDRF (Brisbane, Calgary, Manchester, Toronto) and one by a Michigan Diabetes Research and Training Center grant (Michigan P30DK020572). Some aspects of the Toronto project were funded by a Diabetes Canada (formerly the Canadian Diabetes Association) operating grant; other aspects of the Manchester projects have received funding from NIH (5RO1-NS46259-03, R01DK077903-01A1 NINDS). We acknowledge the contributions of the Steven and Ofra Menkes Fund for supporting aspects of this research in Toronto. The funding sources played no role in study design, in the collection, analysis and interpretation of data, in the writing of the report or in the decision to submit the paper for publication.

Data availability

Data are available from the corresponding author on reasonable request.

Duality of interest

BAP reports grants from CIHR, NIH and JDRF, speaker honoraria from Medtronic, Johnson & Johnson, Insulet, Abbott, Novo Nordisk and Sanofi and research grant support from Medtronic and Boehringer Ingelheim and serves as a consultant for Boehringer Ingelheim, Insulet and Novo Nordisk. NE reports grants from JDRF and NHMRC during the conduct of the study. DP reports grants from NIH RFA DK-13-027 and grants from JDRF during the conduct of the study. JKM has received research grants from Eli Lilly, Pfizer, Bristol-Myers Squibb, PTC Therapeutics, Sanofi Genzyme, Novartis, ReveraGen Pharma and Biogen. SIL reports grants from NIH-DHHS-US through a consortium with Mount Sinai Hospital and grants from NIH-NIDDK Michigan Diabetes Research Center Diabetes Interdisciplinary Study Grant during the conduct of the study. All other authors declare that there is no duality of interest associated with their contribution to this manuscript.

Footnotes

Bruce A. Perkins and Leif E. Lovblom are co-primary authors.

Electronic supplementary material

The online version of this article (10.1007/s00125-018-4653-8) contains peer-reviewed but unedited supplementary material, which is available to authorised users.

Contributor Information

Bruce A. Perkins, Email: ac.metsyshtlaehianis@snikrep.ecurb.

Rayaz A. Malik, Email: ku.ca.retsehcnam@kilam.a.zayar.

References

1. Pop-Busui R, Boulton AJ, Feldman EL, et al. Diabetic neuropathy: a position statement by the American Diabetes Association. Diabetes Care. 2017;40:136–154. doi: 10.2337/dc16-2042. [PMC free article] [PubMed] [CrossRef] [Google Scholar]

2. DCCT Research Group Effect of intensive diabetes treatment on nerve conduction in the Diabetes Control and Complications Trial. Ann Neurol. 1995;38:869–880. doi: 10.1002/ana.410380607. [PubMed] [CrossRef] [Google Scholar]

3. Breiner A, Lovblom LE, Perkins BA, Bril V. Does the prevailing hypothesis that small-fiber dysfunction precedes large-fiber dysfunction apply to type 1 diabetic patients? Diabetes Care. 2014;37:1418–1424. doi: 10.2337/dc13-2005. [PubMed] [CrossRef] [Google Scholar]

4. Chen X, Graham J, Dabbah MA, Petropoulos IN, Tavakoli M, Malik RA. An automatic tool for quantification of nerve fibers in corneal confocal microscopy images. IEEE Trans Biomed Eng. 2017;64:786–794. doi: 10.1109/TBME.2016.2573642. [PMC free article] [PubMed] [CrossRef] [Google Scholar]

5. Tavakoli M, Quattrini C, Abbott C, et al. Corneal confocal microscopy: a novel noninvasive test to diagnose and stratify the severity of human diabetic neuropathy. Diabetes Care. 2010;33:1792–1797. doi: 10.2337/dc10-0253. [PMC free article] [PubMed] [CrossRef] [Google Scholar]

6. Ahmed A, Bril V, Orszag A, et al. Detection of diabetic sensorimotor polyneuropathy by corneal confocal microscopy in type 1 diabetes: a concurrent validity study. Diabetes Care. 2012;35:821–828. doi: 10.2337/dc11-1396. [PMC free article] [PubMed] [CrossRef] [Google Scholar]

7. Jiang MS, Yuan Y, Gu ZX, Zhuang SL. Corneal confocal microscopy for assessment of diabetic peripheral neuropathy: a meta-analysis. Br J Ophthalmol. 2016;100:9–14. doi: 10.1136/bjophthalmol-2014-306038. [PubMed] [CrossRef] [Google Scholar]

8. Bossuyt PM, Reitsma JB, Bruns DE, et al. STARD 2015: an updated list of essential items for reporting diagnostic accuracy studies. BMJ. 2015;351:h5527. doi: 10.1136/bmj.h5527. [PMC free article] [PubMed] [CrossRef] [Google Scholar]

9. Tavakoli M, Malik RA. Corneal confocal microscopy: a novel non-invasive technique to quantify small fibre pathology in peripheral neuropathies. J Vis Exp. 2011;47:2194–3001. [PMC free article] [PubMed] [Google Scholar]

10. Tavakoli M, Ferdousi M, Petropoulos IN, et al. Normative values for corneal nerve morphology assessed using corneal confocal microscopy: a multinational normative data set. Diabetes Care. 2015;38:838–843. doi: 10.2337/dc14-2311. [PMC free article] [PubMed] [CrossRef] [Google Scholar]

Articles from Diabetologia are provided here courtesy of Springer