Subcutaneous Insulin Administration: Sufficient Progress or Ongoing Need?

Millions of people with diabetes around the globe practice subcutaneous (SC) insulin administration (SIA) every day.¹ The significant progress made in insulin needle technology over the last 10 to 20 years has considerably improved insulin injection. With modern short, narrow-diameter, thin-walled cannula, sharper needle tips, and silicone lubrication, insulin administration from syringes or pens is far less painful in comparison to the situation when insulin was discovered.^2,3 Likewise, introduction of continuous subcutaneous insulin infusion (CSII), has enabled better glycemic control, especially when paired with recent automated insulin delivery (AID) systems.⁴ However, many SIA unmet needs remain, including delivering insulin in the most minimally invasive and physiologically relevant manner.^5,6 Characteristics of “ideal” SIA would include

• - completely free from side effects

• - pain free

• - highly reproducible and predictable

• - low cost

Research About SIA and Call for Increased Action

Considering an almost 100-year history of insulin usage, the ubiquity of the SC route for insulin administration, increasing diabetes prevalence, and intense focus on tighter control using advanced algorithms and decision support systems, one might think that all aspects of SIA are a hot topic with extensive ongoing research. Interestingly the opposite appears to be the case, with few new contributions to SIA appearing at scientific meetings or in the literature. This is demonstrated by a brief survey analysis of SIA publications and patents over time (Figure 1). It appears that publications on this topic have begun to level off and even decrease in recent years, despite a considerable increase in the number of patents issued per year in the area. From our point of view, this reflects that background technical knowledge around insulin delivery and absorption from the SC depot into the blood stream has remained static, despite the increased number of patents on the topic.

Figure 1.

Publications and patents found in literature and US patent database search about SIA.

PubMed search query: (“injections, subcutaneous” [MeSH Terms]) AND “insulin” [MeSH Terms] OR (subcutaneous*[Title/Abstract] AND insulin[Title/Abstract]) AND (injection[Title/Abstract] OR administration[Title/Abstract]).

Patent search query: CTB=(subcutaneous* AND insulin) AND CTB=(administration OR injection) AND DP>=(18360101).

Another interpretation of this literature analysis is that is already sufficient understanding of SIA, with little to be gained from further investigation of this topic. However, despite the availability of rapid-acting insulin analogs (RIA) for a number of years, many—if not the majority of—patients with diabetes still fail to achieve optimal postprandial glucose (PPG) control.⁷ This elusive goal highlights the issues around SIA consistency and predictability. Increased knowledge about SIA might also assist development of novel ultrafast insulins (UFI) and predictive control algorithms.⁸ Improved understanding of factors influencing insulin absorption could achieve more physiologic and predictable time-action profiles, leading to better PPG control. We believe there is a significant need to expand the body of SIA knowledge and improve our intrinsic understanding of the SIA process including the mechanics and systems for delivery, tissue perfusion and uptake, and local and systemic factors that impact PK and PD (pharmacokinetics and pharmacodynamics) of insulin.

Factors Influencing SIA

SIA is known to have both substantial intra- and intersubject variability in both insulin uptake and metabolic control.⁹ It is generally accepted that the rate of insulin absorption, from injection or infusion, depends on multiple factors, which can be biological, mechanical, formulation-based, or interconnected in nature. Intrinsic biological factors known to impact absorption such as tissue type (intradermal, SC, or intramuscular), dispersion of insulin depots, and degree of local blood flow can also in turn be affected by mechanical or formulation influences. For example, increased local blood flow affects SIA by increasing the diffusion gradient between the insulin depot and the insulin concentration in the blood traveling through the capillaries, that is, more insulin will diffuse across the capillary walls into the blood stream. Increased local blood flow can be achieved by warming the skin, rubbing the injection site, or by injecting formulation additives that increase insulin distribution.¹⁰ Similarly, formulation excipients may impart a range of effects including keeping insulin molecules as monomers, increasing local blood flow, or increasing insulin distribution in tissue. It is of interest to note that most attempts to develop UFI using these excipient options in the last 10 years have been unsuccessful, with only one marketed product, Fiasp, using excipients to increase local blood flow, although other UFI formulation strategies are undergoing clinical trials. Undoubtedly, an increased understanding of these interacting factors to improve consistency of effect could have significant impact on diabetes therapy.

Needle-Based SIA

Insulin administration into the SC tissue is still predominantly accomplished with pen needles; studies have shown decreases in pain with new shorter, narrower, and sharper pen needle technology.^2,3 Advantages of shorter, smaller gauge needles include less pain, due to either physical or psychological perception, lowered risk of intramuscular injection, and a simpler technique with no pinch-up requirement.^2,11 To the inexperienced eye it might appear simple to design a new needle or CSII set (for insulin infusion by means of pumps); however, these represent high technological hurdles, particularly when considering manufacturing at the consistency, volume, and quality needed to serve an extensive patient population. Similarly, an optimal SIA needle-based injection system should ideally accommodate the vast differences in user techniques (eg, pinch up or not, pen grip, applied device force), ergonomics, and site selection commonly encountered during real-world patient usage.

Form of the Insulin Depot

The shape of the insulin depot as deposited in the SC tissue is not spherical as commonly presumed. Animal studies have shown that insulin depot exhibits broad variance, depending on the anatomical structures in the SC tissue, with insulin predominantly distributing in channels between adipocytes.¹² Such differences in shape and thereby surface area of the depot—even between equivalent insulin doses applied at the same injection site—might explain why insulin absorption is so highly variable. Significant limitations of this study include that insulin delivery was into deceased swine tissue and was infused rather than injected, potentially biasing the results considering the different rates of delivery.

It is clear that the distribution in the SC tissue has an impact on insulin absorption. “Insulin spreading” results in faster absorption, demonstrated by a recent study in humans showing that nine injections of two units acted faster than a single injection of 18 units.¹³ In another study, use of hyaluronidase as an additive to enhance bulk fluid flow (ie, “spreading”) in the SC space, both faster insulin absorption and reduced variability in insulin absorption and action were demonstrated.¹⁴ The more rapid absorption and reduced variability seen in these studies were likely a product of larger surface area: volume ratio. While the insulin depot is an important factor in insulin distribution and absorption, the articles referenced here, although notable, have only been cited <20 times since their publication in the years 2011 and 2013.

In a recent unpublished clinical study by one author (RJP), clinical MRI has been used to examine depot formation of saline injected via pen needle in the abdominal SC tissue of normal subjects.¹⁵ Results demonstrate the irregular dispersion of injection depots in living human tissue, and depot sizes proportional to the injected volume (Figure 2). Hopefully, advanced imaging methodology such as this can be used to better correlate clinical PK/PD outcomes to the varied effects of insulin deposition in future prospective clinical studies.

Figure 2.

Reconstructed SC depots of 60 (green), 30 (blue), and 10U (pink) volume equivalent of saline superimposed over underlying abdominal muscle tissue (magenta) from high resolution MRI. Tissue depots exhibit irregular deposition patterns, and volumes correlated to injectate volume. Source: Pettis and Rini, unpublished data.

SIA Into Lipohypertrophic Tissue

One unwanted side effect of SIA is tissue reactions against repeated injections or infusions of insulin into the same skin area. The root causes for the development of lipohypertrophy (LHT) or lipoatrophy are not fully understood; possible mechanisms include immune reaction to insulin or its formulation excipients for lipoatrophy,¹⁶ and to repetitive injection trauma and / or the anabolic effects of insulin itself in the case of LHT. One main reason for these unwanted side effects of insulin therapy is incorrect injection technique, such as chronic use of the same injection site (incorrectly rotating injection sites), and reuse of needles, especially more than 4 or 5 times.^17-20 Additional nonmodifiable risk factors include duration of insulin usage and more frequent daily injections. The type of insulin, volume or dose of insulin injection, type of needle, BMI, and age are not considered risk factors for LHT. Nearly all studies show that patients with LHT take higher TDD (total daily dose) of insulin, usually associated with worse glycemic control (higher HbA1c). However, this is a result of injecting into LHT, and not a causal factor.^19-21

This known complication of insulin therapy shows up in a surprisingly high number of patients, with the reported prevalence varying from 16 to 60%.²¹ More recent studies generally put the prevalence at least 50%. One reason for this broad range of reported frequencies is that LHT diagnostic procedures are not well standardized and the outcome depends on the skill of the clinical examiner. There is a need for a quick and reproducible method for diagnosis of LHT. Visualization of LHT via thermography was evaluated in one small pilot study, but showed limited sensitivity and specificity, suggesting it is not likely to become widely used in this regard.²² Ultrasound has been proposed to diagnose LHT, but there is no agreed-upon standard for the appearance of LHT lesions vs normal SC adipose tissue, to date.

The clinical relevance of SIA into LHT has been confirmed with a recent rigorously conducted euglycemic clamp study demonstrating ~20-40% reduced insulin absorption after SIA into SC areas with LHT, as well as a 3-5× worsening of within-subject variability (CV%) of insulin uptake, compared to injecting into normal adipose tissue.²³ Additionally, a standardized meal study also showed that LHT again reduced insulin uptake, and not surprisingly, led to clinically meaningful deterioration of PPG control.²⁴ Despite these clear findings, there is limited research to reduce the prevalence of this common SIA side effect beyond the current training recommendations for promoting effective delivery site rotation and avoiding needle reuse.^25,26

Recent interventional trials have evaluated the efficacy of proper injection technique training to improve glycemic control and reduce glycemic variability, in both subjects with and without preexisting LHT.^24,27,28 While these studies have been generally positive, one was uncontrolled and of short duration, and one was a pilot trial; the third showed positive effects but there was considerable contamination or “wash-in” of the control group of patients. Definitive proof of the cause-and-effect relationship between poor injection practice, LHT, and glycemic control is still lacking.

Local Blood Flow in the SC Tissue

Local blood flow is affected by numerous physiological and patient-specific factors, including age, BMI and skin area, leading to region-specific differences in insulin uptake. In addition, environmental factors like temperature or other drugs might also impact local blood flow. A number of methods and devices are commercially available that allow measurement of blood flow in the skin; however, these do not allow measurement of blood flow in the SC tissue. The correlation between blood flow in both compartments is not well understood, this is also due to concerns about the validity of methods proposed for measurement of blood flow in the SC tissue. In view of the importance of this factor on insulin absorption, it is of surprise that no validated technique is available.

Two potential techniques might be useful: laser Doppler ultrasound and optical coherence tomography. The latter is depth limited to the skin itself and shallow SC tissue. Doppler ultrasound likely requires a contrast media. Laser Doppler measurements of local skin blood flow have been shown to increase along with temperature.²⁹ This increased blood flow results in faster absorption of insulin, as has been shown in studies with a medical product that warms up the skin in the area of the SC insulin depot.³⁰ It has to be evaluated if magnetic resonance could provide a better insight into blood flow in the SC tissue; however, this technique requires specific expertise and expensive hardware.

Absorption of Insulin via the Lymphatic System and Effects of Tissue Microenvironment

Until recently, insulin was believed to be almost exclusively absorbed via the blood stream, that is, to be more precise, that insulin dimers and monomers dissociate from hexamer form, and the monomers diffuse from the insulin depot through the SC interstitial matrix to be absorbed through the capillary vascular endothelium into the blood stream. This belief is based on a number of historical studies with radio-labeled insulin that were performed some 30 to 40 years ago (many in Denmark).^31-33 However, recent studies performed with swine and humans in which dyes³⁴ or insulin^35-37 were applied intradermally (ID) have shown much more rapid transfer of insulin via the lymphatic system than was previously reported. Additionally, multicompartment input models have been shown to have better fit to actual clinical PK data, implying parallel uptake pathways.^38,39 This implies that insulin absorption via the lymph may be an underappreciated aspect of insulin uptake. In addition, vascular and lymphatic capillary densities within the tissue, molecular diffusion rates through the extracellular matrix, dissociation rates of various insulins and their specific formulation dependence, and local cellular utilization or degradation may also all contribute to the total uptake during SIA.⁴⁰ Clearly, a more comprehensive approach and increased research focus on local tissue microenvironment would lead to better understanding and control of the SIA process.

Summary and Outlook

The aim of this editorial is to generate more interest in SIA and serve as a call to action. Improvement of our knowledge and more interest about the details of SIA and insulin transport from the SC into the blood stream will help further improve insulin therapy in patients with diabetes. A systematic and comprehensive evaluation of factors influencing SIA and basic research to achieve a better understanding of the details of SIA is needed, ideally from a company independent perspective. Various approaches may be used to speed up SIA to have even better UFI in the future.

Clearly, research has a lot to do with funding, that is, if academic sites / research facilities would like to be more active in this area, respective funding organizations (like NIH) should value such patient oriented research more highly. Industrial players in diabetes tend to invest into research in proportion to the market size for a given business segment and return on their research investment, therefore funding for core mechanistic understanding of SIA might be below that of developing new products. While understandable from a business perspective, this limitation may be shortsighted from the broader viewpoint of the comprehensive diabetes community. Ways to facilitate such vital research needs via a consortia-based approach are warranted.

In view of the digital revolution that our world is undergoing, the impact of “smart” products for insulin applications (eg, for insulin pens, dosing algorithms, carbohydrate calculators) is also just beginning to be felt. To have precise and accurate information about the timing and dose of insulin delivery, along with better assessment of carbohydrate intake, will also be a big step forward toward AID systems without insulin pumps.