Specific skin signs as a cutaneous marker of diabetes mellitus and the prediabetic state – a systematic review

Cutaneous manifestations are seen in a minimum of 30% of diabetics during the course of their illness, and some may even appear before the diagnosis of diabetes mellitus (DM) is confirmed [1-4]. Besides being markers of DM, skin signs may possibly play a central role in reducing the complications associated with diabetes as they may improve the motivation of patients and physicians towards disease management [5, 6]. Epidemiological data strongly suggest that type 2 diabetes mellitus (T2DM) is expanding at an epidemic rate; and it has been estimated that the number of diabetics will reach 250-300 million globally by 2025 [7].

The aim of this study was to determine whether skin signs are feasible as cutaneous markers for the prediabetic state as well as overt DM.

METHODS

A systematic search was conducted to identify any specific cutaneous manifestations of DM (Figure 1A). For this purpose, the databases PubMed, Embase and Cochrane were used. The search strategy is shown in Figure 1B. The search was conducted in accordance with the PRISMA guidelines and following the PICO model [8], and the final search date was 5 November 2015. We excluded studies of populations with confounding conditions like malignancies, thyroiditis, gestational diabetes and treatment with drugs affecting the skin. We also excluded animal studies, studies with poor methodology, pilot studies and studies focusing on more than one

cutaneous manifestation. The included articles were screened for relevant cross-references. The selection process resulted in a total of 34 articles (Table 1).

RESULTS

Acanthosis nigricans

Acanthosis nigricans (AN) was first recognised in 1889 by a German dermatologist [9]. Initially, AN was thought to appear among obese and overweight individuals, but it was later clarified that the condition was linked to increased levels of insulin, which stimulates keratinocytes and dermal fibroblasts through insulin-like growth factor (IGF) [6, 10, 11]. As a consequence, thickened and hyperpigmented velvety skin develops in body folds, mostly the axillae and flexural areas of the posterior neck (Figure 2A and Figure 2B). The literature supports a higher frequency of AN among the darker-skin coloured populations [6, 10-13]. It is more difficult to detect AN in fair-skinned people, where the condition may go undetected until overt DM develops. Mostly, the posterior neck and intertriginous areas are involved, but it can also be observed on the knuckles or elbows. AN is often seen in cases of hyperinsulinaemia, and therefore it may also be considered a cutaneous manifestation of the prediabetic state [11, 13]. Studies repeatedly find a

higher frequency of AN among females [12-15], and one study found it to be an independent risk factor for insulin resistance in overweight Hispanic children [16]. Weight-loss along with increased physical activity alleviates the condition by reducing the circulating insulin

level [6, 17]. Assigning the diagnosis of AN and emphasising its relation to insulin resistance and possible ways to improve the condition may motivate patients to improve their lifestyle [11]. Screening for AN is a non-invasive procedure, and suggested optimal screening is initiated at the age of five and repeated every second year for individuals with multiple risk factors such as a family history of T2DM, ethnicity and obesity [13]. It is also important to keep in mind that there are other aetiologies to AN, including malignancies [6, 10].

Skin tags

These common skin lesions were first described in relation to DM in 1951 [18]. Skin tags (ST), also named acrochordons, appear as small and soft skin tumours which are often located on the lateral aspects of the neck, back, axilla, trunk and face (Figure 2B) [19]. Of the fibroepithelial tumours of the skin, ST are considered the most common [20]. Hyperinsulinaemia stimulates the IGF-I receptor, which has a tissue proliferative effect resulting in ST.

Therefore, individuals with ST are observed to have hyperinsulinaemia and glycated haemoglobin (HbA1c) levels at the high end of the normal range [21]. There are significantly more ST cases in the diabetic population than among normoglycaemic individuals [20]. As with AN, the ST population presents with a significantly

higher body mass index (BMI), especially when their medical history indicates the presence of additional diabetogenic risk factors [19, 20]. A positive correlation has been found between the total number of ST and mean fasting plasma glucose. Patients with more than 30 ST are at an increased risk of developing DM, and females with ST beneath their breasts are particularly prone to developing DM [19]. It seems that ST may have a higher prognostic value for DM than obesity [19]. It is important to stress that hypertension, hyperlipidaemia, atherosclerosis and AN appear at a much higher rate in the ST population [21-23].

Diabetic dermopathy

The condition was first described by the Swedish physician Hans Melin in 1964 [24] and named diabetic dermopathy (DD) by Binkley in 1965 [25]. This manifestation, also named shin spots, is the most common cutaneous manifestation of DM and it is almost considered pathognomonic [26]. It has been reported in up to 55% of patients, most frequently in males [27-30]. DD is observed on the pretibial skin as asymmetric, bilateral atrophic and hyperpigmented round-to-oval macules and is diagnosed clinically (Figure 2C). Studies suggest that DD has a microangiopathic origin [27], which should alert physicians to investigate for other diabetic complications related to the microvasculature including retinopathy and peripheral neuropathy. Additionally, neuropathy may be an aetiologic factor and it has been postulated that patients with peripheral neuropathy are more prone to trauma, leading to the development of DD lesions [29]. The reduced pain sensation may explain why individuals with these lesions tend to neglect the presence their lesions. DD has been reported to be associated with a number of other diabetic complications, which warrants further investigation [27]. Although DD is said to be pathognomonic for DM, it should be kept in mind that prolonged intake of certain medications such as iron-containing drugs, antimalarials and quinolones may lead to lesions imitating DD [27, 31].

Rubeosis faciei

First described in the early 19th century as a characteristic reddening of the face (Figure 2D) [32], this manifestation is particularly prevalent among diabetic Ashkenazi Jews [33]. In a recent study, the prevalence of rubeosis faciei (RF) was found to be 7% in patients with type 1

diabetes mellitus (T1DM) [34]. The condition can be exacerbated by hypertension, and the underlying mechanism is suggested to be diabetic microangiopathy, which results in dilatation of superficial veins. The degree of RF reflects the incidence of DM, and also the severity of microangiopathy [33]. As with ST, many patients with RF might go undiagnosed.

Pruritus

This manifestation presents itself as localised or generalised itching without visible primary skin lesions [35]. Generalised pruritus (PR) has been reported in up to 50% of patients with DM [36, 37] and can be an initial symptom of DM [38]. Many diabetics have involvement of the genital and truncal areas [37, 39]. There seems to be a correlation between postprandial glucose and generalised PR in patients with DM; however, no relationship between HbA1c levels and generalised itching has been found [37]. Diabetic smokers are predisposed to developing PR to a higher degree than non-diabetic smokers [35]. Additionally, PR is a marker for polyneuropathy as it appears among patients who lack the Achilles tendon reflex or patients who present with numbness of their soles and palms [39].

Granuloma annulare

Colcott Fox was the first to describe granuloma annulare (GA) in 1895 [40]. GA may be generalised or localised, and the generalised distribution seems to be more prevalent in diabetics [41-43]. The skin lesions of GA are typically symmetrically distributed along the distal region of the extremities and sun-exposed skin areas [41]. The papular eruption gradually expands with skin-coloured or red borders, creating a central involuted crater (Figure 2E and Figure 2F). GA was found to be more prevalent among adult females [43].

The similarities with necrobiosis lipoidica (NL) are striking, why many have hypothesised that GA is seen prior to this skin disorder [43]. Nevertheless, several studies have failed to show an association between GA and DM [44-46]. Even so, it has been recommended that all individuals presenting with GA should be screened for DM, and it was also suggested that individuals above the age of 30 years with generalised GA should be screened annually [42].

Necrobiosis lipoidica

This condition was originally described by Erich Urbach in 1932 [47]. As mentioned above, NL can resemble GA and is commonly associated with DM [43]. The pathogenesis is not clear, but may be explained by microangiopathic changes and hypoxia [48-50], although an antibody-mediated vasculitis has also been suggested [51]. The lesions exhibit the Koebner phenomenon and are usually painless although some patients may experience itching or pain, especially if the skin lesions ulcerate [52]. Studies from the 1960s showed that more than 60% of patients with NL had DM and 42% of NL patients were undiagnosed diabetics [52, 53]. Subsequent studies, however, have found a less obvious correlation [54]. The typical lesion is a sharply defined atrophic telangiectatic plaque with a glazed content of yellow-brown colour. NL is most commonly located pretibially (Figure 2G), and in 80% of cases it will have a bilateral distribution. Usually, diabetics develop lesions in their fourth decade of life [54]. The prevalence of NL among diabetics is 0.3%, and the condition is more frequent among women. Forty-three percent of the patients who have this condition have a family history of DM [53, 54].

Scleroedema diabeticorum

The first full clinical description of this scleroederma-like disease is attributed to Buschke in 1902 [55]. The condition manifests by severe permanent thickening of the skin of the posterior neck and upper back. Typically, the skin thickening develops over years and presents like peau d’orange, resulting in decreased sensation to pain and touch in the affected areas (Figure 2I) [56, 57]. Scleroedema diabeticorum (SD) is primarily seen in individuals with long-standing DM. The prevalence of SD among T2DM patients is approximately 2.5% [57]. On the other hand, more than 90% of those with this lesion have DM. A suggested mechanism is the effect of hyperglycaemia on collagen in the skin, where skin biopsies of diabetics show skin thickening with swollen collagen [56]. One consequence of the thickened skin is a reduced range of motion especially affecting the upper back, shoulder and posterior neck [56, 57]. Others are particularly affected on their hands and fingers [58]. In these patients, the condition can easily be demonstrated by the “prayer sign”, where patients are asked to oppose the two hands firmly (Figure 2J) [59, 60]. It is important to distinguish SD from scleroedema Buschke, even though some may use the terms interchangeably. Scleroedema Buschke is usually used to describe a post-infectious condition, which is self-limiting [56].

Bullosis diabeticorum

The formation of bullae in diabetics was first observed in 1930 [61]. The condition is not inflammatory, but may relate to neuropathy or vascular insufficiency [62]. Bullosis diabeticorum (BD) is described as resembling a blister secondary to a burn, without the erythematous surrounding. The condition usually appears spontaneously overnight, excluding minor trauma as the causative mechanism, and the bullae vary in size from 0.5-10 cm2 containing sterile fluid (Figure 2K and Figure 2L) [62-64]. BD is considered a benign lesion with total remission within 5-10 weeks. During the healing process, the fluid leaks leaving a darkened crust over the affected area which dissolves spontaneously. Most frequently, it is observed as a unilateral lesion involving the foot. Nonetheless, bilateral affections as well as multiple lesions have been reported, and other regions such as the trunk, arms and hands can be affected. The prevalence is 0.16% in tertiary care centres, and there is a higher prevalence among males with a median age of 65 years [62-64]. Typically, patients have longstanding T1DM. However, it has also been recognised as a prediabetic manifestation. This rare lesion should be identified by physicians and lead to appropriate intervention to avoid further complications. There is a risk of developing secondary infection, which may complicate the condition, leading to a prolonged healing time [65]. Patients often describe the onset of symptoms as a burning sensation that later fades away. Recurrence is often observed. It is important to exclude other bullous entities, making BD a diagnosis of exclusion.

DISCUSSION

Skins signs are observed among diabetics with variable duration and also in the prediabetic state. AN is often seen early in the course of the disease, especially when an individual has additional risk factors for developing DM, including obesity. The majority of included studies use HbA1c, fasting plasma glucose measurements and homeostatic model assessment. Regardless of the

methods applied, there is overall agreement that AN is significantly associated with DM [10, 12, 16]. Kobaissi and coworkers diverted from the other studies by applying a computer programme (MINMOD) to the obtained blood samples to define insulin sensitivity [16]. They found a stronger correlation with obesity, and they therefore suggested that obesity should be a marker for DM. Detection of AN has also been shown to encourage discussion between the patient and the clinician leading to earlier intervention [5]. Screening is recommended due to its inexpensive and non-invasive nature [11, 16].

ST and AN are frequently found among diabetics and prediabetics and can co-occur. An Iranian case-control study conducted among 104 patients with ST and 94 BMI-matched controls showed a higher frequency of DM in patients with ST [19]. The study found a significant correlation between 30 or more ST and risk of impaired carbohydrate metabolism. Akpinar and coworkers did not manage to reproduce a statistically significant difference between patients with ST and controls using fasting plasma glucose levels [20]. The discrepancy may be due to the fact that Rasi and coworkers included participants with 3 ST or more in their cross-sectional study making them more likely to document significant differences. Demir & Demir found a significant correlation of ST with hyperinsulinaemia rather than overt DM [22].

We identified no original studies focusing on DD and its relation to DM, which may be taken to underscore the fact that the manifestation is acknowledged as pathognomic for DM. Studies on DD lack consistency in their methodological approach and clarification of the underlying pathophysiology [27, 28, 31]. All the diabetics with dermopathic lesions in the included studies showed underlying microangiopathy. Thus, an indirect correlation can be deduced from the included studies, as well as DD’s function as a marker for diabetic complications resulting from microangiopathy.

Manifestations like RF are easy to recognise when the patient is first seen. While prevalence studies show a tendency for RF to occur among diabetics, evidence of a direct correlation is still lacking. Skin colour, ambient temperature, hormonal factors and other skin diseases such as rosacea may influence facial redness. Further studies are required to elucidate the pathophysiology of this phenomenon.

PR is an unspecific symptom and could be related to other causes including, fungal infections. In a Taiwanese study, postprandial glucose was found to be significantly elevated among patients with generalised PR. To the best of our knowledge, this is the only study to date reporting on this relation. In contrast, in a cross-sectional study from the UK including 100 diabetics with PR [37], no correlation between generalised PR and DM could be demonstrated. However, there was a positive correlation with vulvar PR. Both studies had shortcomings in relation to testing for autonomic neuropathy, which may influence perspiration and the perception of sensory signals. A matter of concern is the difference in terminology and the methods used in the two studies. Even though there are disputes regarding the correlation of PR to DM, it would be beneficial for physicians to include DM among their differential diagnoses when presented with a patient having unexplained generalised PR. A thorough investigation of any diabetic person presenting with PR is advised as this diffuse symptom is also seen with other diseases such as thyroid disorders and lymphoma [66].

There is also a dispute regarding the relation between GA and DM as recent studies have been unable to reproduce the results shown by earlier studies [46].

The reason for the inconsistency between the results achieved in the studies may be that different definitions and investigations of DM were used as some used fasting plasma glucose and others used HbA1c. Another point of interest is the variable phenotypical presentation of GA. It seems to be the generalised form that is linked to DM [41, 45]. In summary, there is an insufficient amount of studies to properly support any statements regarding the degree of correlation between GA and DM.

NL has been linked to DM in earlier studies [53, 54]. However, in later studies, participants with NL did not suffer from DM. The study by O’Toole et al therefore raises the question whether NL is a cutaneous manifestation “of the past” [54]. However, physicians should still be aware of this manifestation and its occurrence among diabetics. There seems to be no recent studies on this subject.

SD is rarely seen nowadays. It is mainly reported among obese diabetics with poorly controlled DM sampled from highly selected populations, e.g. diabetes clinics [56, 57]. Thus, the reported prevalence may be higher than among the general diabetic population. Nonetheless, the collected data support the notion that SD develops from prolonged exposure of high blood glucose on collagen in the skin leading to thickening of the skin.

In contrast to some of the other manifestations, BD is specifically reported among diabetics, indicating a strong clinical correlation. The occurrence of BD among diabetics is possibly under-recognised as all publications on this subject are seemingly case reports or smaller case series. A critique of some of these reports is that many of the subjects present with comorbidities that may contribute to the development of bullae.

Besides these classically described diabetes markers, there is increasing knowledge of a correlation between psoriasis and T2DM [67]. A systematic review and meta-analysis from 2013 suggests an association between diabetes and psoriasis and psoriatic arthritis. It should be noted that the heterogeneity of the included studies raises questions with respect to extrapolation of their results. Nonetheless, the evidence to date indicates that physicians should screen patients with psoriasis for diabetes [68].

Strengths and limitations

To the best of our knowledge, this is the first systematic review aiming to investigate the role of the different skin lesions as markers for the prediabetic state or overt DM. The literature search presents an updated picture of the available evidence and reveals some issues that require further investigation.

A general critique, which applies to many of the studies, is sampling bias. Thus, many of the participants were selected from diabetic clinics or other tertiary units. Another major critique is the lack of blinding of physicians and participants which may both entail further bias. An additional important difference is the use of inconsistent and incompatible methods. Even though current studies are likely to provide an updated view, it is important to keep in mind that the older studies better reflect the poorly controlled diabetics.

CONCLUSION

The most recent studies demonstrate a strong association between DM and the entities of ST and AN. The remaining presented manifestations require further investigation inasmuch as newer studies cannot reproduce previous results. However, lack of evidence should not prevent physicians from screening patients, as this is a low-cost measure and as early detection is beneficial. The management of DM has improved markedly over the years, and better glycaemic control results in a reduction of complications.

Recognition of cutaneous markers enables an earlier diagnosis of undiagnosed DM and recognition of suboptimal management of known disease. Thus, the presented skin sigsns should lead to a DM-focused diagnostic evaluation.

Correspondence: Rewend Salman Bustan. E-mail: rsa-5@hotmail.com

Accepted: 27 October 2016

Conflicts of interest: Disclosure forms provided by the authors are available with the full text of this article at www.danmedj.dk