Abstract
Background:
Many patients with diabetes on insulin therapy develop lipohypertrophies (LHTs). So far, LHTs are diagnosed by conventional methods (CM; visual inspection, palpation and/or ultrasound). In everyday life, it would be advantageous to have a quick, simple and inexpensive alternative, for example, diagnosing them by obtaining infrared (IR) images.
Methods:
We obtained IR images from 43 subjects (21 patients with type 1 diabetes, conventional subcutaneous insulin therapy and known LHTs, 8 patients with CSII and LHTs, 7 patients without LHTs, and 7 healthy people), all from one specialized diabetes practice. The IR images were taken under standardized conditions with a high-resolution infrared camera (VarioCam® HDx Jenoptic, IR pixel 640 × 480, thermal resolution 0.003K) and compared with LHT diagnoses with CM.
Results:
In 14 of the 29 (48%) patients, CM diagnosed LHTs were “cold spots” in the IR images. The temperature difference to “healthy” skin (without LHTs) was up to 6°C. Of the 14 patients, 11 also showed such spots, without findings with CM. Four patients did not show clearly identifiable cold spots as LHT and 2 patients showed no changes in the IR images. The remaining 9 patients did not show clearly identifiable cold spots as LHT, but the diagnosis with CM was also ambiguous.
Conclusions:
The results of this small (pilot) study do not clearly support the value of IR images for the diagnosis of LHTs, but they do not refute this approach. Diagnosis of LHT might be hampered due to the existence of different types of LHTs. Usage of IR images can apparently detect LHTs before they can be diagnosed with CM. Further targeted investigations are required to make statements about the usability of this method.
Keywords: lipohypertrophy, diabetes mellitus, infrared, thermography, insulin therapy
According to some reports at least one-third of all patients with diabetes on insulin therapy develop lipohypertrophic (LHT) skin changes during the course of their disease.1 Beside the cosmetic harm, it is long-standing clinical experience that subcutaneous injection into LHTs impairs insulin absorption. The extent of such deteriorations was confirmed recently by clinical-experimental pharmacokinetic and pharmacodynamics studies, documenting the worsening of postprandial glycemic control if insulin is applied to LHTs.2,3
Up to now in daily practice three methods (conventional methods, CM) are used most often for detection of LHT: inspection, palpation, and ultrasound.4 An experienced examiner can detect LHTs with CM with a sufficient reliability; however, in view of the large number of patients that probably have LHTs, a more rapid and easy to apply method to visualize LHTs would be of tremendous help in clinical practice. Facing the clinical introduction of closed loop systems in the near future, pathologies extending the absorption process in the subcutaneous tissue are of special interest. Reducing the barrier for LHT detection might also help to bring LHTs to the attention of physicians and diabetes nurses In addition, training of patients would be easier when LHTs can be visualized more easily. Better training in injection technique in turn most probably helps to avoid development of LHTs.
Thermography, that is, measurement of infrared radiation from the human body, is a diagnostic option which is in medical use for a range of application, for example, plastic surgery to determine the blood supply of skin regions to construct cutaneous flaps.5
We evaluated if usage of infrared (IR) images represents a painless, fast and easy to apply option for diagnosis of LHT in patients with diabetes.
Materials and Methods
Subjects
For this evaluation patients with type 1 diabetes (T1D) on different types of insulin therapy and with and without known existence of LHTs were recruited; healthy subjects acted as a control group. All subjects were recruited in one specialized diabetes practice in Leverkusen, North Rhine Westphalia, Germany. The patients with known LHTs resemble the most distinct cases with this diabetes therapy related complication in this practice.
In total 43 subjects participated in this evaluation study: 21 are patients with T1D on a conventional intensified insulin therapy (MDI), that is, with subcutaneous insulin therapy by means of insulin pens, eight patients used a continuous subcutaneous insulin infusion therapy (CSII), seven patients had no (known) LHTs and seven were healthy subjects (Table 1). Approval of the local ethics committee and written informed consent from all subjects was obtained prior to the study.
Table 1.
Demographic and Clinical Characteristics of Study Participants.
| All patients (n = 43) | MDI (n = 21) | CSII (n = 8) | Control; T1 or T2 with insulin but without LHTs (n = 7) | Control; healthy subjects (HS) (n = 7) | |
|---|---|---|---|---|---|
| Female (%) | 33 | 38 | 13 | 29 | 86 |
| Age (years) | 54 (15) | 59 (12) | 49 (13) | 58 (15) | 37 (17) |
| BMI (kg/m2) | 27.9 (5.4) | 29.3 (5.2) | 26.8 (4.7) | 30.5 (5.8) | 22.4 (2.5) |
| Insulin therapy (years) | 19 (10) | 29 (8) | 23 (22) |
Values are mean (SD) unless otherwise noted.
Preparation
To establish comparable measurement conditions, prior to the examination all study participants should follow these rules: They should have no sunbed, sauna, steam bath, body lotion, powder or engage in sportive exercise within 24 hours before the IR images were taken. They should also avoid the consumptions of nicotine, bubble gum, and food within three hours prior to the examination. Compliance with these rules, as well as a room temperature of 22°C were checked before every examination.
Material and Specification
To obtain high quality IR images a camera with a high resolution and sensitivity was used (VarioCam® HDx Jenoptic; head 675/20 mm, IR-Pixel 640 × 480, thermal resolution 0.003K). For the analysis of the stored IR images a specialized software was used (IRBIS® 3). When not in use, the IR camera was stored in the examination room and calibrated before every usage.
Measurements
During the measurements the IR camera was placed a 1 m distance and an angle of 0° above the abdomen of the lying and undressed subject (Figure 1). During a “cool down” period of 10 minutes, IR images are taken every second. At the end of this period, single images of the abdomen from different angles were made to obtain optimal coverage of the abdomen. Afterward a visual inspection of the abdomen was performed, existence of LHT was also checked by palpation and ultrasound by an experienced diabetologist (MK). For the standardization of the results, they were mapped on a dimensional sketch.
Figure 1.
Experiment setup to obtain infrared images (with an image of a healthy control subject).
Endpoints
The primary endpoint of this evaluation was the ability to diagnose LHT by usage of IR images in comparison to that diagnosed by CM. Secondary endpoints were practical experience with an IR camera and handling aspects as well as technical requirements.
Results
The IR images of the abdomen of the healthy subjects showed considerable individual variations (Figures 2 and 3a). The IR images of the abdomen of patients with LHT showed temperature differences between +6°C and −6°C (“hot” and “cold” spots). During the acclimatization period of 10 minutes, no structural change of the heat radiated by the abdomen was observed; however, the contrast between hot and cold areas increased.
Figure 2.
Infrared images of healthy subjects with no history of diabetes (ages 19, 58).
Figure 3.
Infrared image of the abdomen with cold spots (LHTs confirmed by CM) of a healthy subject from the control group (a) and of a patient with T1D treated with MDI for 25 years (b).
LHTs diagnosed by CM on the abdomen of patients with T1D corresponds with distinct areas in the IR images, these are colder structures (“cold spots”) (Figures 3b and 4). Four of the seven healthy subjects showed—as to be expected—no sign of LHT in the IR images (Figure 2). However, the other three subjects—that were older than the other subjects—showed suspicious spots (Figure 3a). These probably physiologic spots could be misinterpreted as LHTs. Two of the seven subjects of the control group with patients with T1D with no LHTs diagnosed with CM had no indication of a LHT in the IR images. The remaining five subjects showed suspicious spots that appear like LHTs.
Figure 4.
Infrared image and photos of the LHTs in the lower part of abdomen with prominent LHTs of a patient with T1D treated with MDI since 17 years.
In 14 of the 29 (48%) patients, CM diagnosed LHTs were “cold spots” in the IR images (Figure 4). Of the 14 patients, 11 also showed such spots, without findings with CM (Figure 5). Four patients did not show clearly identifiable cold spots as LHT and 2 patients showed no changes in the IR images (Figure 6). The remaining 9 patients did not show clearly identifiable cold spots as LHT, but the diagnosis with CM was also ambiguous. So, the correlation between LHTs diagnosed with CM and those seen in IR images showed a good agreement in general (Figure 7, Table 2) and many, but not all individual cases (Figure 6).
Figure 5.
Infrared image and photo of the abdomen of a patient with T1D treated with MDI since 13 years. Diagnosis of LHTs with CM (blue-marked skin areas) and IR image match well. However, the IR image reveals more suspect areas with LHTs. After experiencing several hypoglycemic events when injecting insulin into the lower abdomen, patient is currently preferring the upper abdomen for insulin administration.
Figure 6.
Infrared image and photos of the abdomen of patient with T1D treated with CSII in the last 8 years. The LHT, clearly visible in the photos, is not prominent in the IR image.
Figure 7.
Correlation between diagnoses of LHT by means of conventional methods versus usage of infrared images (see also Table 2).
Table 2.
Results of Usage of Conventional Methods or Infrared Images for the Diagnosis of LHT for the Individual Patients (a) and Summarized (b).
| (a) | ||
|---|---|---|
| Patient number | Conventional methods Explicitness 1 = nothing (–) 2 = questionable (+–) 3 = obvious (+) |
Infrared images Explicitness 1 = nothing (–) 2 = questionable (+–) 3 = obvious (+) |
| MDI-1 | 3 | 2 |
| MDI-2 | 3 | 2 |
| MDI-3 | 3 | 3 |
| MDI-4 | 3 | 3 |
| MDI-5 | 3 | 2 |
| MDI-6 | 2 | 3 |
| MDI-7 | 2 | 2 |
| MDI-8 | 3 | 3 |
| MDI-9 | 2 | 2 |
| MDI-10 | 2 | 2 |
| MDI-11 | 3 | 3 |
| MDI-12 | 3 | 3 |
| MDI-13 | 2 | 2 |
| MDI-14 | 3 | 3 |
| MDI-15 | 3 | 3 |
| MDI-16 | 3 | 2 |
| MDI-17 | 3 | 3 |
| MDI-18 | 3 | 3 |
| MDI-19 | 3 | 1 |
| MDI-20 | 3 | 3 |
| MDI-21 | 2 | 2 |
| CSII-1 | 2 | 2 |
| CSII-2 | 2 | 2 |
| CSII-3 | 2 | 3 |
| CSII-4 | 2 | 2 |
| CSII-5 | 3 | 3 |
| CSII-6 | 2 | 2 |
| CSII-7 | 3 | 3 |
| CSII-8 | 3 | 1 |
| Control-T1-1 | 1 | 2 |
| Control-T1-2 | 1 | 1 |
| Control-T1-3 | 1 | 2 |
| Control-T2-4 | 1 | 2 |
| Control-T2-5 | 1 | 2 |
| Control-T2-6 | 1 | 2 |
| Control-T2-7 | 1 | 1 |
| Control-HS-1 | 1 | 2 |
| Control-HS-2 | 1 | 1 |
| Control-HS-3 | 1 | 1 |
| Control-HS-4 | 1 | 2 |
| Control-HS-5 | 1 | 1 |
| Control-HS-6 | 1 | 2 |
| Control-HS-7 | 1 | 1 |
| (b) | ||
| Number of patients | Conventional methods Explicitness 1 = nothing (–) 2 = questionable (+–) 3 = obvious (+) |
Infrared images Explicitness 1 = nothing (–) 2 = questionable (+–) 3 = obvious (+) |
| 6 | 1 | 1 |
| 8 | 1 | 2 |
| 9 | 2 | 2 |
| 2 | 2 | 3 |
| 2 | 3 | 1 |
| 4 | 3 | 2 |
| 12 | 3 | 3 |
Discussion
The results of this small (pilot) study—which aimed to evaluate if IR images can be used for diagnosis and disclosure of LHTs—are not straightforward. In about half of the cases, LHTs diagnosed with CM (that have also a number of limitations) showed clear association with cold spots in respective areas in the IR images; however, in the other half, an obvious LHT could sometimes not be detected in the IR images or vice versa. Thus, we cannot recommend the use of IR images for the diagnosis of LHT in daily clinical practice; however, we believe that further studies are required before abandoning this approach completely.
Explanations why in some patients the colder spots in the IR images did or did not correspond to LHTs diagnosed by CM might be that such LHTs are too small to be diagnosed by CM and/or that different types of LHT exist (with different blood flow/heat).
Assuming that the development of LHTs in the skin is a dynamic (and reversible) process, IR images might tell a different story about LHT in comparison to CM. This consideration is based on reports by patients about a relatively fast normalization of the skin when changing their insulin administration scheme. In contrast, others report to have LHT since years, despite not using them as sides of injection. The consideration of different “types” of LHTs, that also might have different IR images, is in line with reports by others.6 It appears as if LHT have not one clear definable pathology, but a number of different kinds of LHTs exist.6 Some develop rapidly and the remission takes not a long time, others grow slowly and persist for years. The microscopic processes within the tissue and the exact causes of the development of these transformations in the subcutaneous tissue is not well explored.
Our measurements also indicate some kind of inhomogeneity of LHTs, it appears as if LHTs in the IR images are individual, that is, there is no “standard” image. However, also the limited insight into the interpretation of IR images has to be mentioned. In the same line of thinking the methods used might not be standardized/reproducible enough. Nevertheless, as mentioned before, also CM has difficulties to diagnose LHTs.
A factor that might be responsible for hot spots in the IR images is an ongoing inflammation rising the temperature in a given skin area. A lowered blood flow, probably due to fibrovascular processes/hypertrophic adipocytes in LHTs might be the reason for cold spots in the IR images.
An interesting (and not easy to interpret) observation is spots in the IR images of control group patients that can be interpreted as LHTs. Also the individual variations are of note; however, due to the small sample size no firm statements can be made why this is the case. It appears as if the heat radiated by the human body is influenced by age, weight, and sex. The colder spots around the umbilicus and on skinfolds in all subjects might represent “physiologic” colder spots; these seem to increase with age and weight and appear to be more symmetrically arranged.
Also the thickness of the subcutaneous fat layer (or the amount of fat deposited inside the abdomen) might have an effect on the surface temperature as measured by the IR camera. In other words, the IR images reflects the temperature at the skin level, but this might be driven in a complex interaction by a number of other factors. An irregular distribution of adipose tissue in the abdomen, with a predominant spot around the umbilicus, might therefore be a challenge for the interpretation of IR images. Asymmetrically ordered colder spots in the IR images are more likely to be related to LHT. Colder spots around the umbilicus, that are ordered symmetrically, on the other hand are more likely to be physiological, but as LHTs often appear at the same location, this might be the cause for false positive results in the control group and could be responsible for difficulties in insulin therapy, despite the presence of LHTs, if these findings are of pharmacokinetic relevance.
It is of interest to note that during the “run-in” phase no changes in the IR images could be observed, that is, cold spot remained at the same location and did not disappear (change in local blood flow). However, the contrast between different skin areas increases, that is, temperature differences became more pronounced.
Costs of a diagnostic method are an important aspect for daily usage. For this evaluation study—which aimed to proof the concept of using IR images as a method to detect LHTs—an expensive camera with a high technical performance (an excellent resolution and high thermal resolution) was used. Initially an inexpensive camera was used that can be attached to a smart phone (with a resolution of 80 × 60 pixels and a thermal resolution of 0.15 K). The IR images obtained did by far not provide the information gained by the expensive device; from the perspective of costs, an intermediate device might fulfill the requirements. It is worth mentioning that this is a one-time investment only; no additional costs are associated. The time required for taking an IR image is clearly shorter in contrast to a sound ultrasound examination.
In view of the high clinical relevance of LHT we see the need to explore the usability of IR images for diagnosis of LHT further. An advantage of using this method can be that beside the diagnosis of LHT that can be diagnosed rapidly by CM, IR images enables visualization of skin areas with LHTs that could not (yet) be detected by CM. If this observation can be confirmed, this might have clinical relevance as patients apply their insulin in presumably “healthy” skin areas without LHT, while such skin changes already exist and change insulin absorption properties. It would be of interest to apply insulin in LHTs detected by IR images but not by CM in glucose clamp studies and evaluate the impact of doing so on pharmacokinetic/pharmacodynamic properties of the applied insulin formulation. Previous studies have shown a correlation between the surface temperature and the absorption properties of insulin; however, these studies are limited by the fact that they were performed only at the same location on the abdomen.7
In view of the constantly rising number of patients with diabetes on insulin therapy, the risk of developing LHTs and their negative influence on a successful insulin therapy also increases. Thus, there is the need for further research looking into the option of using IR images for LHT diagnosis, a safe, painless, rapid, relatively low cost, and easy to apply technology. Further studies might also enable to make clearer statement regarding the sensitivity and the specificity of using IR images for diagnosis of LHTs. They might also help to understand which level of measurement quality is required for a reliable LHT diagnosis with IR images in clinical routine. As rotating sites of subcutaneous insulin injection in the abdominal region is the most efficient way to prevent the development of LHTs, IR images might have a great impact on patients awareness and training of appropriate changes in the practical aspects of their insulin therapy.8,9 Proper injection site rotation means that successive injections should be moved by roughly 1 finger width (1 cm) from the previous injection.10 Moving injections from one side of the abdomen to the other, back and forth, is not sufficient, as this will lead to two areas of LHT if the injections are performed in the same place, time after time.
An “advantage” of LHT is for the patients that injecting the insulin into such skin areas is associated with less pain.9 Changes in insulin absorption properties are not directly obvious for them. A comparable diagnostic and educational issue with patients is the diabetes foot syndrome.11 As thermography plays a rising role in this pathology as well, the further use of IR images in education efforts for patients with diabetes is advisable.
Acknowledgments
We thank Dietmar Weber for his thorough reading of the manuscript.
Footnotes
Abbreviations: CM, conventional methods; CSII, continuous subcutaneous insulin infusion; IR, infrared; LHT, lipohypertrophies.
Declaration of Conflicting Interests: The author(s) declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: Becton-Dickinson Germany supported lending of the infrared camera for this evaluation study by means of an unrestricted grant.
Funding: The author(s) received no financial support for the research, authorship, and/or publication of this article.
References
- 1. Deng N, Zhang X, Zhao F, Wang Y, He H. Prevalence of lipohypertrophy in insulin-treated diabetes patients: a systematic review and meta-analysis. J Diabetes Investig. 2018;9(3):536-543. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Famulla S, Hövelmann U, Fischer A, et al. Insulin injection into lipohypertrophic tissue: blunted and more variable insulin absorption and action and impaired postprandial glucose control. Diabetes Care. 2016;39:1486-1492. [DOI] [PubMed] [Google Scholar]
- 3. Ji L, Sun Z, Li Q, et al. Lipohypertrophy in China: prevalence, risk factors, insulin consumption, and clinical impact. Diabetes Technol Ther. 2017;19:61-67. [DOI] [PubMed] [Google Scholar]
- 4. Hambridge K. The management of lipohypertrophy in diabetes care. Br J Nurs. 2007;16:520-524. [DOI] [PubMed] [Google Scholar]
- 5. John HE, Niumsawatt V, Rozen WM, Whitaker IS. Clinical applications of dynamic infrared thermography in plastic surgery: a systematic review. Gland Surg. 2016;5:122-132. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Bertuzzi F, Meneghini E, Bruschi E, Luzi L, Nichelatti M, Epis O. Ultrasound characterization of insulin induced lipohypertrophy in type 1 diabetes mellitus. J Endocrinol Invest. 2017;40:1107-1113. [DOI] [PubMed] [Google Scholar]
- 7. Sindelka G, Heinemann L, Berger M, Frenck W, Chantelau E. Effect of insulin concentration, subcutaneous fat thickness and skin temperature on subcutaneous insulin absorption in healthy subjects. Diabetologia. 1994;37:377-380. [DOI] [PubMed] [Google Scholar]
- 8. Campinos C, Le Floch JP, Petit C, et al. An effective intervention for diabetic lipohypertrophy: results of a randomized, controlled, prospective multicenter study in France. Diabetes Technol Ther. 2017;19(11):623-632. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Frid AH, Hirsch LJ, Menchior AR, Morel DR, Strauss KW. Worldwide injection technique questionnaire study: injecting complications and the role of the professional. Mayo Clin Proc. 2016;91:1224-1230. [DOI] [PubMed] [Google Scholar]
- 10. Frid AH, Kreugel G, Grassi G, et al. New insulin delivery recommendations. Mayo Clin Proc. 2016;91:1231-1255. [DOI] [PubMed] [Google Scholar]
- 11. Adam M, Ng EYK, Tan JH, Heng ML, Tong JWK, Acharya UR. Computer aided diagnosis of diabetic foot using infrared thermography: a review. Comput Biol Med. 2017;91:326-336. [DOI] [PubMed] [Google Scholar]