Open access Original research Increased plasma neurofilament light chain levels in patients with type-1 diabetes with impaired awareness of hypoglycemia Frederic Sampedro,1 Nicole Stantonyonge,2,3 Saül Martínez-­Horta,1 Daniel Alcolea,1 Alberto Lleó,1 Laia Muñoz,1 Rocío Pérez-­González,1 Juan Marín-­Lahoz,1 Beatriz Gómez-­Ansón,4 Ana Chico ‍ ‍2 To cite: Sampedro F, Abstract Stantonyonge N, Introduction Impaired awareness of hypoglycemia (IAH) Significance of this study Martínez-­Horta S, et al. is a common complication in patients with type-1 diabetes Increased plasma (T1D). IAH is a major risk factor for severe hypoglycemic What is already known about this subject? neurofilament light chain ►► Patients with type 1 diabetes (T1D) with impaired events, leading to adverse clinical consequences and levels in patients with type- awareness of hypoglycemia (IAH) are in higher risk 1 diabetes with impaired cerebral damage. Non-­invasive, cost-­effective, and logistically efficient biomarkers for this condition have not of severe hypoglycemia (SH). awareness of hypoglycemia. been validated. Here, we propose plasma neurofilament ►► SH is associated with adverse clinical consequences BMJ Open Diab Res Care light chain (NfL) levels as a biomarker of neuroaxonal and cerebral damage. 2020;8:e001516. doi:10.1136/ bmjdrc-2020-001516 damage in patients with T1D-­IAH. ►► There is no validated biomarker for this subgroup of Research design and methods 54 patients were patients. included into the study (18 T1D-­IAH, 18 T1D with normal What are the new findings? FS and NS contributed equally. awareness of hypoglycemia (NAH) and 18 healthy ►► Plasma neurofilament light chain (NfL) levels were Received 4 May 2020 controls). We measured plasma NfL levels and studied increased in patients with T1D-­IAH compared to pa- Revised 23 May 2020 cerebral gray matter alterations on MRI. tients with T1D with normal awareness of hypogly- Accepted 30 May 2020 Results We found that NfL levels were increased in cemia and healthy controls. patients with T1D-­IAH compared with patients with T1D-­ ►► Increased NfL levels correlated with reduced cere- NAH and healthy controls. Importantly, increased NfL levels bral gray matter volume and increased IAH severity correlated with reduced cerebral gray matter volume and in patients with T1D-­IAH. increased IAH severity in patients with T1D-­IAH. Conclusion Overall, our findings identify plasma NfL How might these results change the focus of levels as a potential biomarker of cerebral damage in this research or clinical practice? population, motivating further confirmatory studies with ►► Plasma Nfl levels could be used as a non-­invasive, potential implications in clinical trials. cost-­effective, and logistically efficient biomarker of © Author(s) (or their neuroaxonal damage in patients with T1D-­IAH. employer(s)) 2020. Re-­use permitted under CC BY-­NC. No Introduction commercial re-­use. See rights About one in five patients with type 1 diabetes and permissions. Published (T1D) have impaired awareness of hypogly- logistics, standardization, scalability, and by BMJ. cemia (IAH).1 IAH represents a risk factor for cost-­ efficiency. On the other hand, while 1 Neurology, Hospital de severe and recurrent hypoglycemic events, cerebrospinal fluid biomarkers of cerebral la Santa Creu i Sant Pau, which can lead to brain damage.2 3 As no effec- damage overcome some of these limitations, Barcelona, Spain 2 Endocrinology, Hospital de tive treatment has been clearly established,4 they require an invasive lumbar puncture la Santa Creu i Sant Pau, biomarkers capable of detecting and quanti- procedure. Barcelona, Spain fying cerebral damage in this population are In recent years, research has focused 3 School of Medicine, needed for the design of novel preventive or on the detection of blood biomarkers of Autonomous University therapeutic strategies. neural damage. Clearly, cost-­efficient blood of Barcelona, Barcelona, Catalunya, Spain On the one hand, cerebral atrophy or biomarkers would represent an accessible 4 Neuroradiology, Hospital functional impairment can be assessed using and centralizable alternative. However, the de la Santa Creu i Sant Pau, neuroimaging techniques. In fact, a struc- main challenge for the development of an Barcelona, Spain tural and functional brain compromise has adequate technology in this context has been Correspondence to been described in patients with T1D-­ IAH.5 the low blood concentration of proteins Dr Ana Chico; However, neuroimaging techniques have derived from the central nervous system ​achicob@​santpau.​cat important limitations in terms of specificity, (CNS). Importantly, in the last few years, the BMJ Open Diab Res Care 2020;8:e001516. doi:10.1136/bmjdrc-2020-001516 1 Pathophysiology/Complications Single Molecule Array (Simoa) technology has shown analogs). Sociodemographic, clinical, and laboratory unprecedented precision and sensitivity (ie, femtomolar data (diabetes and cardiovascular indicators) were range) for the quantification of blood biomarkers of brain recorded for each patient. Quantification of IAH severity damage.6 In particular, increased levels of neurofilament was assessed, on one hand, by Clarke scores. On the other light chain (NfL) in blood have shown a great potential hand, we also recorded the number of severe hypogly- for the detection of neurodegeneration (neuronal loss cemic episodes during the last 5 years and the number of and axonal damage), traumatic brain injury, and cere- moderate hypoglycemic events during the last month. A brovascular accidents.7 severe hypoglycemic episode associated loss of conscious- In this brief report, we investigate whether patients ness, confusion/disorientation, seizures, or coma and with T1D with hypoglycemia unawareness show increased required the assistance of a third person. A moderate levels of plasma NfL compared with patients with normal hypoglycemic event was defined as any non-­ severe awareness of hypoglycemia (NAH) and healthy controls episode requiring the use of measures with impact on (HC). Additionally, we explore whether abnormally the daily life (interrupting work routine, stopping the car increased NfL levels in this population correlate with while driving, and so on). We combined both indicators IAH severity and reduced cerebral gray matter volume to obtain a single indicator of the number of severe or (GMV). moderate hypoglycemic episodes, providing a represen- tative indicator of the clinical consequences of IAH. Materials and methods Plasma NfL procedure Sample and assessments Blood was collected in EDTA tubes, centrifuged at 2000 g A total of 54 participants were included in this study: for 10 min and the plasma aliquoted in polypropylene 36 patients with T1D and 18 HC. Eighteen patients tubes, and frozen at −80°C according to international with T1D fulfilled clinical criteria for IAH (T1D-­IAH) consensus recommendations until analysis. Plasma NfL and the other 18 patients had NAH (T1D-­NAH). Both levels were measured with the Simoa Human NF-­light T1D groups were selected as to have similar sociodemo- Advantage kit using the Single Molecule Array tech- graphics and T1D duration. The Clarke score was used for nology (Simoa; Quanterix)9 in a SR-­X Biomarker detec- the IAH classification, which is based on eight questions tion system by following the manufacturer’s instructions. that allow the determination of the glycemic threshold in which the patients notice the hypoglycemic symptoms.8 A Clarke score ≥4 was used to identify patients with T1D-­ Neuroimaging acquisition and preprocessing IAH, whereas patients with T1D-­NAH showed a score <3. T1-­weighted MRI scans were acquired in a Philips 3-­Tesla The Clarke method comprises the following set of ques- station using a magnetization-­prepared rapid acquisition tions designed to characterize the participant’s exposure with gradient echo protocol using the following param- to hypoglycemia: Do you feel hypoglycemia symptoms eters: flip angle 8°, repetition time 6.7 s, echo time 3.19 when your blood glucose is low?, Have you lost symptoms ms, pixel bandwidth 241, slice thickness 1.2 mm, and of hypoglycemia that used to occur when your blood an acquisition matrix of size 256×256. All patients had glucose is low?, Check all your symptoms of hypoglycemia available T1-­MRI imaging, but nine subjects from the HC over the last 6 months, If your answered ‘yes’ to any of group did not have MRI data. the above question, how often do you have it?, Check all GMVs of the frontal, temporal, parietal, and occip- your symptoms of hypoglycemia over the last 12 months. ital lobes for each participant were computed using the If your answered ‘yes’ to any of the above question, how FreeSurfer 6.0 software package. Details about the stan- often do you have it? How often did you have an episode dard FreeSurfer segmentation pipeline are available of hypoglycemia with symptoms during the last month? elsewhere (https://​surfer.​nmr.​mgh.​harvard.​edu/). All How often did you have an episode of hypoglycemia preprocessed images were visually inspected to confirm without symptoms during the last month? What was the the absence of segmentation errors. lowest blood glucose level before feeling the symptoms of hypoglycemia? How often did you have symptoms of Statistical analyses hypoglycemia when your blood glucose level was low? Clinical, sociodemographics, NfL levels, and imaging The following exclusion criteria were applied: previous data were compared across groups using one-­way ANOVA clinical diagnosis of cognitive impairment or neurolog- (except χ2 for sex differences). Pairwise group differ- ical disorders, use of medications affecting the CNS, ences were examined by general linear models including and MRI contraindications or intolerance. None of the age, sex, education, and T1D duration as covariates of no participants in this study had any previous medical condi- interest. A logistic regression was also performed. In addi- tion (apart from T1D) known to significantly alter blood tion, we explored whether increased plasma NfL levels NfL levels or cerebral GMV. correlated with increased IAH severity or reduced GMV Both T1D groups were under similar insulin treat- in patients with T1D. For that, we used one-­sided Pear- ment, either as basal-­ bolus regimen or continuous son’s correlation coefficients using the same nuisance subcutaneous insulin infusion (with short-­acting insulin covariates, for which a p<0.05 was considered significant. 2 BMJ Open Diab Res Care 2020;8:e001516. doi:10.1136/bmjdrc-2020-001516 Pathophysiology/Complications Table 1 Clinical, demographic, biomarker, and imaging information across groups Significance T1D-­IAH T1D-­NAH HC (p value) n 18 18 18 Age (years) 55.5±13.5 57.4±8.1 54.7±13.7 0.79 Sex (% female) 61% 27% 44% 0.13 Education (years) 14.8±5.1 14.7±4.4 13.6±3.5 0.64 Plasma NfL (pg/mL) 18.22±11.7 13.27±6.7 8.15±4.6 0.002* T1D duration 31.7±12.8 30.2±8.6 NA 0.66 HbA1c (%) 7.57±1.1 7.60±0.8 NA 0.93 Body mass index (kg/m2) 25.9±6.6 27.2±3.8 NA 0.49 Mean arterial pressure (mm Hg) 93.6±15.6 101.8±13.6 NA 0.10 LDL-­cholesterol (mg/dL) 96.3±27.4 94.6±21.3 NA 0.84 HDL-­cholesterol (mg/dL) 71.4±16.5 63.7±19.1 NA 0.21 Clarke score 4.7±0.9 1.0±0.7 NA <0.001* Number of severe or moderate hypoglycemic episodes 4.89±6.3 0.61±1.0 NA 0.007* Minimental state examination 29.56±0.6 29.61±0.8 NA 0.81 Frontal lobe GMV (mm3×10–3) 138.6±9.9 150.3±13.5 165.0±17.6 <0.001* Temporal lobe GMV (mm3×10–3) 93.0±7.2 102.3±9.7 100.8±9.6 0.007* 3 –3 Parietal lobe GMV (mm ×10 ) 89.6±8.7 97.3±9.4 108.2±8.8 <0.001* Occipital lobe GMV (mm3×10–3) 41.7±4.8 48.3±6.8 45.1±6.6 0.008* Values are expressed as mean±SD or percentage. *P<0.05 three-­group ANOVA/χ2 or t-­test between T1D groups. ANOVA, analysis of variance; GMV, gray matter volumes; HC, healthy controls; HDL, high-­density lipoprotein; IAH, impaired awareness of hypoglycemia; LDL, low-­density lipoprotein; NA, not applicable or available; NAH, normal awareness of hypoglycemia; NfL, neurofilament light chain; T1D, type-1 diabetes. Results Finally, the following exploratory associations between Table 1 shows the sample characteristics of this study. All increased NfL levels in patients with T1D and IAH severity groups had similar sociodemographics. There were no and reduced cerebral GMV were observed. Within the significant differences between T1D groups in terms of T1D-­ IAH group, increased NfL levels correlated with disease duration, cognitive status, or cardiovascular risk lower frontal GMV (r=−0.51, p=0.030) and increased factors. However, as expected, patients with T1D-­ IAH number of severe or moderate hypoglycemic events showed increased Clarke scores and number of severe (r=0.49, p=0.038). When considering all patients with or moderate hypoglycemic events in the last month, T1D, NfL levels correlated with Clarke scores (r=0.34, reflecting common clinical consequences of IAH. p=0.029) and the number of severe or moderate hypo- Significant differences were found across groups in glycemic events (r=0.45, p=0.005). Finally, although plasma NfL, progressively increasing from HC to T1D-­ non-­ significant, NfL levels showed a mild association NAH and T1D-­IAH. Pairwise NfL differences between with temporal and parietal GMV (r=−0.28, p=0.051 and T1D-­IAH and T1D-­NAH were significant after controlling r=−0.21, p=0.11, respectively). for age, sex, education, and disease duration (β=0.039, Figure 1 displays NfL differences across groups and a p=0.038). A logistic regression to classify patients with subset of these associations, using in some cases log NfL T1D-­IAH versus T1-­NAH including all factors listed in values to better illustrate group differences or correla- table 1 (except for Clarke scores and number of hypo- tions, for which all statistical results remained significant. glycemic episodes) yielded the following results. Signifi- cance was only achieved by NfL levels (p=0.038), but the following close-­to-­significance factors were also observed: Discussion frontal GMV (p=0.062), temporal GMV (p=0.059), age Our results reinforce the presence of increased brain (p=0.066), sex (p=0.064), and body mass index (p=0.054). damage in patients with T1D-­IAH. Plasma NfL levels were Similarly, progressively decreasing GMV was observed increased in this population and correlated with reduced across groups and cerebral lobules, and pairwise differ- cortical gray matter and IAH severity. Therefore, plasma ences between T1D-­IAH and T1D-­NAH remained signif- NfL could be a promising cost-­ efficient biomarker of icant after controlling for the same nuisance covariates. neural damage in patients with T1D-­IAH. BMJ Open Diab Res Care 2020;8:e001516. doi:10.1136/bmjdrc-2020-001516 3 Pathophysiology/Complications Figure 1 (A) Plasma NfL differences across groups (*p<0.05, **p<0.005). Scatter plots illustrating a subset of significant associations between NfL levels and (B) Clarke scores in patients with T1D, (C) frontal lobe GMV within the T1D-­IAH group, and (D) number of severe or moderate hypoglycemic episodes within the T1D-­IAH group. GMV, gray matter volumes; HC, healthy controls; IAH, impaired awareness of hypoglycemia; NAH, normal awareness of hypoglycemia; NfL, neurofilament light chain; T1D, type-1 diabetes. Patients with T1D-­IAH showed reduced cerebral GMV, this population, which could be used to predict IAH or especially in the frontal lobe. A frontal compromise has monitor cerebral damage in clinical trials. been previously described in this population, both in neuroimaging and cognitive studies.5 10 Importantly, in Acknowledgements The authors thank Eugenia Mato, CIBER-­BBN, Santa Creu i Sant Pau Hospital, for help with the neurofilament processing. patients with T1D-­IAH, increased NfL levels correlated with reduced frontal GMV. This association shows that Contributors FS wrote the manuscript and contributed to statistical and image analysis. NS researched data, executed the study, and reviewed the manuscript. increased NfL levels reflect cerebral damage in this SMH reviewed the manuscript. DA, AL, LM, and RPG contributed to NfL processing population, reinforcing its role as a potential biomarker. and reviewed/edited the manuscript. JML reviewed/edited the manuscript. BGA Moreover, increased NfL levels were also associated with contributed to image analysis and reviewed/edited manuscript. AC researched increased IAH severity, as measured by Clarke scores and data and reviewed/edited the manuscript. FS, NS, and AC are the guarantors of this work. also by the number of severe or moderate hypoglycemic Funding This work was financially supported by a grant from the SPANISH events. This further emphasizes the clinical significance DIABETES SOCIETY. of increased NfL levels in this population, which probably Competing interests None declared. reflect neural damage due to recurrent hypoglycemia. The main strength of this study is the inclusion of Patient consent for publication Not required. patients with T1D-­IAH and T1D-­NAH with similar T1D Ethics approval All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional research committee duration and sociodemographic and cardiovascular and with the 1964 Helsinki Declaration and its later amendments or comparable profiles. This allowed studying the effect of IAH on NfL ethical standards. Ethical approval for this study was obtained from the Ethics levels and cerebral GMV with high specificity. Limitations Committee of the Hospital de la Santa Creu i Sant Pau (http://www.​recercasantpau.​ of this work include the use of a relatively low sample size, cat/​en/​clinical-​research/​ceic/) on 17 April 2018 at the Hospital de la Santa Creu i Sant Pau, Barcelona, Spain (Code: EC/18/158/5224). All participants provided signed some missing data in the control group, and the explor- informed consent. atory character of the correlations between NfL levels Provenance and peer review Not commissioned; externally peer reviewed. and IAH severity and GMV within the T1D-­IAH group. To conclude, we found that plasma NfL levels are Data availability statement Data are available on reasonable request. increased in patients with T1D-­IAH, probably reflecting Open access This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-­NC 4.0) license, which the presence of neuroaxonal damage with clinical conse- permits others to distribute, remix, adapt, build upon this work non-­commercially, quences. Overall, our findings motivate further confir- and license their derivative works on different terms, provided the original work is matory studies on the potential role of this biomarker in 4 BMJ Open Diab Res Care 2020;8:e001516. doi:10.1136/bmjdrc-2020-001516 Pathophysiology/Complications properly cited, appropriate credit is given, any changes made indicated, and the monitoring in people with type 1 diabetes and impaired awareness use is non-­commercial. See: http://​creativecommons.​org/​licenses/​by-​nc/​4.​0/. of hypoglycaemia. Diabet Med 2018;35:483–90. 5 Bednarik P, Moheet AA, Grohn H, et al. Type 1 diabetes and impaired ORCID iD awareness of hypoglycemia are associated with reduced brain gray Ana Chico http://​orcid.​org/​0000-​0001-​6498-​3411 matter volumes. Front Neurosci 2017;11:529. 6 Khalil M, Teunissen CE, Otto M, et al. Neurofilaments as biomarkers in neurological disorders. Nat Rev Neurol 2018;14:577–89. 7 Gaetani L, Blennow K, Calabresi P, et al. Neurofilament light chain as References a biomarker in neurological disorders. J Neurol Neurosurg Psychiatry 1 Geddes J, Schopman JE, Zammitt NN, et al. Prevalence of impaired 2019;90:870–81. awareness of hypoglycaemia in adults with type 1 diabetes. Diabet Med 2008;25:501–4. 8 Clarke WL, Cox DJ, Gonder-­Frederick LA, et al. Reduced awareness 2 Perros P, Deary IJ, Sellar RJ, et al. Brain abnormalities demonstrated of hypoglycemia in adults with IDDM. A prospective study of by magnetic resonance imaging in adult IDDM patients with and hypoglycemic frequency and associated symptoms. Diabetes Care without a history of recurrent severe hypoglycemia. Diabetes Care 1995;18:517–22. 1997;20:1013–8. 9 Rissin DM, Kan CW, Campbell TG, et al. Single-­Molecule 3 Dave KR, Tamariz J, Desai KM, et al. Recurrent hypoglycemia enzyme-­linked immunosorbent assay detects serum proteins at exacerbates cerebral ischemic damage in streptozotocin-­induced subfemtomolar concentrations. Nat Biotechnol 2010;28:595–9. diabetic rats. Stroke 2011;42:1404–11. 10 Hansen TI, Olsen SE, Haferstrom ECD, et al. Cognitive deficits 4 Reddy M, Jugnee N, El Laboudi A, et al. A randomized controlled associated with impaired awareness of hypoglycaemia in type 1 pilot study of continuous glucose monitoring and flash glucose diabetes. Diabetologia 2017;60:971–9. BMJ Open Diab Res Care 2020;8:e001516. doi:10.1136/bmjdrc-2020-001516 5