Ultrasound classification and grading of lipohypertrophy and its impact on glucose variability in Type 1 Diabetes (the TITANIC Study)
: an exploratory study

Student thesis: Doctoral ThesisDoctor of Philosophy


Lipohypertrophy (LH) is a common insulin injection site problem that occurs with repeated exposure to insulin injections in the subcutaneous tissue. Subcutaneous insulin exposure can increase tissue density with hypertrophy and hyperplasia of adipocytes. The effect of insulin when injected into an LH area can be attenuated, potentially leading to glucose variability. Thereby, potentially increasing the risks of diabetes complications, hypoglycaemia and individual distress. The aim of this study was to explore the association between ultrasonographically characterised LH lesions with time in range (TIR 4 to 10 mmol/L) and glucose variability (GV) in adults with Type 1 diabetes (T1DM). The study also aimed to characterise the LH observed in participants.

The study involved two integrated arms, the glucose variability arm (the GV study) and LH characterisation arm (the LH characterisation study). All participants were recruited from clinics at Guys and St Thomas’ Hospitals. The main screening criteria were: people with T1DM using insulin; with evident GV based on the standard deviation (SD) of the mean glucose from their self-monitored blood glucose downloads (SD³ 4.0mmol/L). In the GV study participants TIR and GV (SD of glucose, coefficient of variation (CV), mean amplitude of glucose excursions (MAGE), continuous overlapping net glycaemic action (CONGA), means of the daily differences (MODD), and mean absolute glucose (MAG)) were assessed in two conditions using blind continuous glucose monitoring (CGM): Condition 1, usual insulin injecting behaviour; and Condition 2, injecting in areas assessed to be free of LH- participants insulin doses were reduced at this stage, to reduce the risk of hypoglycaemia. Additional outcomes included: glycaemic control; insulin satisfaction; diabetes distress; and quality of life. Data were also collected on participant injecting behaviours; and at the end of the study interviews were conducted to explore participants’ views of the study. In the LH characterisation study, the participants injection areas were screened using ultrasound (US) with a blind comparison to physical clinical assessment (digital palpation) following a standard protocol for the assessment and recording the LH. These participants were shown were their LH lesions were and given advise on insulin site management (avoiding LH affected tissue) and insulin doses were reviewed to reduce the risk of potential hypoglycaemia.

A total of 27 participants were enrolled into the GV study, of which 15 completed the study and were included in the analysis. The median age of the completing participants was 32 (IQR, 25-60) years (range 20-71 years), with a median duration of T1DM of 14 (IQR, 10-23) years. In terms of the impact of LH on TIR and GV, one third of those participants demonstrated improvement in TIR (4-10mmol/L) ³10%, the remainder showed limited improvement and one participant showed a reduction in TIR >10% (p=0.02). No significant changes were seen in the GV measures. The findings showed a significant improvement in the median percentage of effective bolus insulin injections (based on the glucose response to the insulin), with an increase of 17% from 69% (IQR, 62-73) to 86% (IQR, 82-93) (p<0.001). In some participants there were large reductions in the total daily insulin dose, in five participants these were 25, 20, 9, 8, 6 units respectively; of the remainder most had minimal dose changes, with two having a modest increase of 3 and 4 units. The LH characterisation study involved the US assessment and digital palpation of 74 participants. The US images showed that the LH was heterogenous in its morphology, with the main presenting features being: LH nodules; and areas of diffuse (dense) subcutaneous tissue. Additional observations included hypoechogenic tissue within nodules which could indicate necrotic tissue; inflammatory changes; and disruption to dermal tissue. The US screening identified 740 LH nodules in the participants, with 304 diffuse areas. The most common areas where LH were observed where in the lower abdominal and thigh areas. LH areas were graded from 1-5 (1 = diffuse LH; 2 = nodules <6mm; 3 = nodules >6mm to <8mm; 4 = >8mm to <10mm; and 5 = >10mm). Palpation was shown to have a moderate level of accuracy (Cohen’s kappa 0.44) in detecting LH compared to US, missing over half of the areas detected via US.

The finding from this study provide some important new insights into the morphology of LH and its impact on clinical management. Overall, the study identified that the interaction between LH and glucose regulation is complex and challenging to study. The current data shows that changing injection sites in a small sample of participants with T1DM led to improvements in TIR for some participants and reductions in insulin doses. The study has also revealed that LH is complex and heterogenous in presentation, and a potential method for grading LH has been presented which could be adopted clinically with further validation. As the study was unable to identify a clear estimation of the association between LH and GV, determining the clinical significance of LH remains an important objective for future studies.
Date of Award1 Jul 2020
Original languageEnglish
Awarding Institution
  • King's College London
SupervisorAngus Forbes (Supervisor)

Cite this