Nailfold capillary density in 140 untreated children with juvenile dermatomyositis: an indicator of disease activity

Background We lack a reliable indicator of disease activity in Juvenile Dermatomyositis (JDM), a rare disease. The goal of this study is to identify the association of nailfold capillary End Row Loop (ERL) loss with disease damage in children with newly diagnosed, untreated JDM. Findings We enrolled 140 untreated JDM and 46 age, race and sex matched healthy controls, ages 2–17. We selected items from the Juvenile Myositis Registry for analysis. Variables include average ERL density of 8 fingers, average capillary pattern, hemorrhages, and clinical and laboratory correlates. Laboratory data includes Myositis Specific Antibodies (MSA), disease activity scores (DAS), Childhood Myositis Assessment Scale (CMAS), and standard clinical serologic data. The reduced mean ERL density is 5.1 ± 1.5/mm for untreated JDM vs 7.9 ± 0.9/mm for healthy controls, p < 0.0001, and is associated with DAS-skin, r = -0.27 p = 0.014, which did not change within the age range tested. Untreated JDM with MSA Tif-1-γ had the lowest ERL density, (p = 0.037); their ERL patterns were primarily “open” and the presence of hemorrhages in the nailfold matrix was associated with dysphagia (p = 0.004). Conclusions Decreased JDM ERL density is associated with increased clinical symptoms; nailfold hemorrhages are associated with dysphagia. Duration of untreated disease symptoms and MSA, modify NFC shape. We speculate nailfold characteristics are useful indicators of disease activity in children with JDM before start of therapy. Supplementary Information The online version contains supplementary material available at 10.1186/s12969-023-00903-x.


Background
Although Juvenile dermatomyositis (JDM) is the most common of the pediatric inflammatory myopathies, it is a rare disease.The annual USA average incidence is 3.2 cases/million children/year white, non-Hispanic, 3.3/ million African Americans, and 2.7/million for Hispanic patients, with an overall girl to boy ratio of 2.3 girls:1 boy, and a mean age at JDM diagnosis of 6.7 for girls and 7.3 for boys [1].Children with JDM have both damage to their vascular system and characteristic inflammatory skin involvement, including periorbital, malar and peripheral erythema, Gottron's papules, and in some cases, calcification (Fig. 1).In addition, they often display dysphagia and proximal muscle damage, accompanied by focal weakness.For example, weakness may be initially manifest as difficulty in climbing stairs or getting in/out of the car, or trouble with swallowing.In addition to the musculoskeletal damage, examination of the capillary loops in the nailbeds of the child with active JDM shows that these capillary loops are both decreased in number, compared with healthy age-matched controls, and that they are misshapen (Fig. 2).The capillary loops may be dilated, shrunken or deleted and free blood-hemorrhages-may be present in the nailfold area as well (Fig. 3).
Over 30 years ago, we created the Ann & Robert H. Lurie Children's Hospital of Chicago Juvenile Myositis Registry and Repository.Our recent RNA sequencing (RNA-Seq) of peripheral blood mononuclear cells, as well as skin and muscle biopsies samples from, untreated JDM, each demonstrated the presence of robust transcriptional activity both at diagnosis and when they appeared to be "clinically inactive" [2].These RNA-Seq data supported our previous serologic evidence of inflammatory disease activity in the apparently clinically quiescent JDM [3].This observation led to the speculation that data pertaining to the decreased density of nailfold capillary (NFC) endrow loop (ERL) may provide a qualtitative estimate of disease activity sufficient to guide to therapy for children with JDM [4,5].

Patient population
Untreated JDM (n = 140), who met diagnosis of JDM [6] 1).Written informed consent was obtained from all legally authorized representatives and assent from those patients aged 12 and older.The children with JDM were diagnosed and seen at Lurie Children's between 1993 and 2021, and had proximal muscle weakness, the characteristic rash, varying degrees of elevated muscle enzymes and, since 2000, a T2 weighted MRI image of involved muscle.Children with overlap syndrome (positive anti-U1 RNP, anti-U2 RNP or anti-PM-Scl), Systemic Lupus or Scleroderma were excluded.

Clinical assessments
Routine clinical laboratory tests were performed in the Lurie Children's diagnostic laboratories, while The Oklahoma Medical Research Foundation assayed the Myositis Specific Antibodies (MSA), using immune precipitation and immunodiffusion [7].Their early MSA data were  was obtained by a pediatric rheumatologist; DAS-total score (DAS-T) (range = 0-20), is derived by adding the muscle evaluation (DAS-M) (range = 0-11) to the skin score (DAS-S) (range = 0-9) [8].A physical therapist obtained The Childhood Muscle Assessment Scale (CMAS) [9].The duration of untreated disease (DUD) is defined as the length of time that the parents/caregiver first noted either new physical signs of JDM or change in activity to the date of the first JDM treatment.

Nailfold capillaroscopy
Since 2012, a digital camera (Nikon Coolpix p6000) equipped with a Dermlite2 ProHR provided standardized NFC images (18x) to generate the data and assess inter-rater reliability, prior to 2012, freeze frame videomicroscopy was utilitized as previously described [5].Figure 2 illustrates the difference seen in healthy control and untreated JDM periungual NFCs.Qualitative measures, such as severity of avascularity, and predominant ERL shape are entered on the NFC work sheet (Table 2).
The main patterns of ERLs are open, undefined, crossed, bushy, branched, and tortuous; the predominant pattern was recorded, Fig. 3.The reproducibility of the method was assessed by two experienced readers who analyzed the images of 49 JDM utilizing Photoshop -see accompanying 3-min video of the method (attachment).

Statistical analysis
The association of a panel of JDM clinical factors with the NFC data was assessed using Pearson's correlation co-efficient, correcting for number of comparisons made utilizing the Bonferonni correction.Standard t-tests were employed on other occasions.The association of the shape of the nailfold capillary ERL with the child's MSA was determined by Chi-square analysis.The statistics were performed in SPSS and figures were generated using Graphpad Prism 9 software.

Inter-rater reliability for NFC analysis
Two trained observers, assessing nailfold data from 49 children with JDM were highly correlated for ERL/mm counts (r = 0.817, p = < 0.0001).
This study was initiated approximately 25 years before information about the specificity of the MSAs were identified [10]; 33% of the early cases of JDM in this study did not have current MSA data.Assessment of 92 JDM with both current MSA and ERL pattern data disclosed that their initial patterns were: 41% open, 38% undefined, 12% crossed, 7% bushy, and 2% tortuous (Fig. 5a).When the JDM were grouped by their MSAs, those with P155/140 had a predominant ERL pattern that was more "open" than the patterns in the other groups-MSA negative, MSA Mi-2 combined with P155/140, and all other MSA (p = 0.03).As Fig. 5b presents, the undefined pattern was associated with the shorter DUD, with a median of 3.6 months in comparison to those with a crossed shape (median of 10.6 months) or an open shape (median 5.8 months), (Kuskal-Wallis test, p = 0.036).The undefined pattern was also associated with the highest ERL capillary density with a median of 5.6/mm in comparison to 3.8/mm for the crossed group and 4.2/mm for the open group (Kruskal-Wallis test, p = 0.002).Furthermore, when DUD was dichotomized into either a short (≤ 3 months) and long (> 3 months) duration, those children having a shorter DUD had more undefined predominant NFC pattern than those with a longer untreated disease

Discussion
Twenty-five percent of our untreated JDM are small children with a short attention span, aged 4 or younger at diagnosis [11], making speed and accuracy essential components of this nailfold capillaroscopy method.The method of obtaining the images described in the accompanying video is both useful and reproducible.When the interrater reliability of trained personnel was obtained for 49 JDM, the two readers' data were correlated (r = 0.817, p = < 0.0001) similar to previous reports of nailfold capillary data for adults with Scleroderma [12].Our data concentrates on ERL number and shape, not rate of capillary blood flow.These data also document that, children with the MSA P155/140 (Tif-1-γ) have a significantly greater loss of ERL when compared with all the other MSAs combined, Fig. 4c. Figure 5a illustrates that the open loop shape is the most prevalent in JDM positive for MSA, P155/140 (Tif-1-γ), while Fig. 5b documents that an undefined pattern is associated with a shorter JDM disease duration.The capillaries are specific targets of the inflammatory process in JDM muscle [13]; capillary injury preceeds muscle fiber damage.With respect to the intestine, ERL density is highly associated with the bioavailability of some drugs used in inflammatory bowel disease-Crohn's disease or ulcerative colitis.JDM with reduction in their nailfold capillary ERL may also have weight loss [14].The vWF:Ag is elevated in 24% of untreated JDM p = 0.006 [15]; serum ICAM-1 is increased as well [16].
NFC morphology in patients with JDM is often indistinguishable from children with Scleroderma, but the presence of "bushy loops" is specific for dermatomyositis.Ongoing investigations on the application of artificial intelligence (AI) for analyses of nailfold capillaroscopy images have promising prospects.Standard AI models (such as the Efficient Net deep neural network architecture) can discriminate between NFCs from patients with JDM and healthy controls [17].We have presented the critical and innovative observation that the nailfold ERL density may reflect gastrointestinal function [18]-which can influence both the optimal type and route of immunosuppressive therapy.The weaknesses of this study include the inherent inter-rater variation.

Conclusion
This report provides comprehensive data at diagnosis for a large group of untreated children with a rare disease, JDM, seen by consistent personnel at a single center.The score sheet provides a simple data collection tool, while the short methods video provides photographic guidance to the collection of these data.We can now conclude that decreased nailfold capillary density is more associated with cutaneous disease activity than muscle involvement at JDM diagnosis.This study documents that nailfold hemorrhages are associated with dysphagia, an often overlooked, but critical dysfunction.Finally, we now have evidence that the ERL shape is modified both by the child's MSA, as well as the duration of their untreated symptoms.

Fig. 1 Fig. 2
Fig. 1 Skin characteristics seen in JDM a dilated eyelid capillaries and telangiectasia, b calcification of the finger, c Gottron's papules of the hand and d erythema of the face

Fig. 3
Fig. 3 Nailfold capillaroscopy photos of the six most frequent patterns that are easily seen in the ERL of untreated children with Juvenile Dermatomyositis: open, crossed, tortuous, bushy, branched, undefined

1 =Table 3 Fig. 4 a
Fig. 4 a Nailfold capillary end row loop (ERL) density in 140 untreated children with JDM and 46 healthy controls.The dotted line is the mean ± 2 standard deviations (shaded area) for controls.The mean ERL for JDM was 5.0 ± 1.5/mm, vs the mean ERL for healthy controls, 8.0 ± 0.9/mm, p < 0.0001.b The association of a range of MSA with nailfold end row loop capillary density in 94 untreated children with JDM.c Comparison of nailfold end row loop capillary density for 51 untreated children with JDM positive for P155/140 MSA with 43 untreated JDM positive for other MSA or MSA negative, * = significance at 0.05.d The lack of association of a positive antinuclear antibody with a significant difference in the density of the nailfold capillary end row loops

Fig. 5 a
Fig. 5 a Distribution of predominant ERL pattern in untreated JDM patients by various MSAs b Untreated children with short DUD have more undefined patterns than long DUD group (Chi-square, p = < 0.0001).c Untreated children presenting with dysphagia have more periungual hemorrhages present in their nailfold capillaries than those not presenting with dysphagia (Chi-square, p = 0.004)

Table 1 Demographics of untreated JDM and healthy controls JDM Healthy Control P-value n = 140 n = 46
first available after 2002, and the P155/140 (Tif-1-γ), MJ (NXP-2), and MDA5 (anti-CADM140) antibodies were reported after 2012.Of the 140 untreated JDM in this study, 94 had current MSAs, including these newly reported antibodies.The Disease Activity Score (DAS)

Table 2
Assessment score sheet for ERL in children with JDM and their controls