Mathematical Model Explores Tissue-specific Insulin Resistance in Adolescent Girls with PCOS

By Lina Sorg

Polycystic ovary syndrome (PCOS) is the most common endocrine disorder in women who are of reproductive age. This hormonal condition—in which the ovaries produce an abnormal amount of male sex hormones called androgens—affects between six and 10 percent of women worldwide and is the leading cause of anovulatory infertility. It is typically diagnosed during puberty, when symptoms such as irregular menses and ovarian cysts first arise. Individuals with PCOS have a greater propensity for both obesity and the development of type 2 diabetes.

Many women with PCOS also exhibit insulin resistance (IR) and hyperinsulinemia, which means that they have elevated levels of insulin in their blood. During a minisymposium presentation at the 2024 SIAM Conference on the Life Sciences, which is currently taking place in Portland, Ore., Cecilia Diniz Behn of the Colorado School of Mines used a modified oral minimal model (OMM) of glucose-insulin dynamics to understand metabolic dysregulation and tissue-specific IR in adolescent girls with PCOS. “There are some specific aspects of the metabolic phenotype that we want to understand,” she said. 

Diniz Behn decided to concentrate on adolescence due to the metabolic challenges that arise during this stage of life. For example, puberty is associated with reduced insulin sensitivity (S_I) and increased insulin secretion. Obesity also affects more than 1/3 of U.S. teenagers, who are simultaneously experiencing impaired glucose tolerance at progressively higher rates. Additionally, adolescents with PCOS are 4.5 times more likely to have metabolic syndrome: a combination of conditions that amplify one’s risk of heart disease, stroke, and diabetes. And in a similar vein, the prevalence of type 2 diabetes is increasing among young people—especially females—and presents more aggressively than in adults. 

At any given point, the body provides a constant source of glucose to the brain. People naturally ingest glucose whenever they eat, but glucose is necessary between meals as well. Therefore, excess glucose within the liver and fat tissues feed the brain during fasting periods. “There’s this balance that has to be maintained when we move between the fasting state to the fed state, when we do have glucose coming from food,” Diniz Behn said.

Insulin is the primary hormone that controls glucose homeostasis (see Figure 1). It promotes the reduction of excess glucose from plasma by affecting both glucose production (by suppressing lipolysis and hepatic glucose release in the liver) and glucose disposal (by helping muscle, fat, and liver cells with absorption). IR occurs when the body’s normal response to insulin is blunted and increased amounts of insulin are required to control blood glucose levels. “An interesting thing about IR is that it’s tissue specific,” Diniz Behn said, adding that muscle IR affects glucose uptake, hepatic IR (of the liver) primarily affects glucose production, and adipose IR (of fat tissue) affects lipolysis. “Being able to understand this tissue-specific IR is a really important goal to help with targeted treatments.”

Diniz Behn explained that common measures of S_I—such as the hyperinsulinemic-euglycemic clamp—focus on steady state behavior. While these static measures do provide insight into certain metabolic features, the interactions between glucose and insulin are dynamic. “We can challenge the system and see the dynamics in action with an oral glucose tolerance test (OGTT),” Diniz Behn said. The physiologic OGTT is administered as a glucose drink and allows practitioners to visualize glucose’s effect on the whole-body system. Diniz Behn utilized an existing oral minimal model (OMM) to identify and interpret tissue-specific IR within OGTT data; the corresponding equations analyze glucose, insulin action, and S_I. Here, insulin acts as a forcing function. “If we take this model and fit it to somebody’s glucose and insulin data, we can describe the trajectory of glucose response,” Diniz Behn said.

She and her team recruited female participants between the ages of 12 and 18 from the PCOS Clinic at the Children’s Hospital Colorado. All participants had a body mass index in the 90th percentile or above, were in the Tanner stage 5 level of puberty, and generally led a sedentary lifestyle. Glucose and insulin levels among the cohort varied drastically. For example, it took six hours after the OGTT for many patients’ blood glucose concentrations to return to baseline; in contrast, people with normal glucose tolerance return to baseline within two to three hours.

In light of this discrepancy, Diniz Behn developed a collection of models of different durations (with and without exponential assumption) and compared the resulting estimates of S_I. “The take-home message is that the models are very highly correlated, but some of the them gave us biased values for insulin sensitivity,” Diniz Behn said. She identified the six-hour S_I model as the gold standard that demonstrated impaired glycemic control in PCOS patients, though limits of agreement suggest that the four-hour truncated model can differentiate S_I between groups and performs almost equally well.

Next, Diniz Behn modeled tissue-specific IR by identifying different sources of glucose production (see Figure 2). Because glucose’s point of origin within the body is not inherently obvious, an OGTT protocol with stable isotope tracers identifies its original source as either ingested food, the liver, or adipose fat tissue. Diniz Behn examined two different tracers: an IV tracer and a drink tracer. “We can then track how the increments change over time and tell which glucose is which in all of the different blood samples,” she said. She used these tracers to indicate which glucose is exogenous (ingested), which is endogenous (liver), and which is the total glucose that reflects both sources.

As with the previous OMM model, insulin acts on glucose disposal via the rate of disappearance. But now it specifically reduces endogenous glucose production. Preliminary results suggested that hepatic S_I accounted for roughly 53 percent of reduced total S_I for participants with PCOS. “This result is consistent with the central role of hepatic S_I in PCOS and higher rates of fatty liver disease,” Diniz Behn said.

Ultimately, metabolic dynamics provide valuable insights into specific features of metabolic function and dysfunction in adolescent females with PCOS. Given reduced hepatic S_I in PCOS patients, Diniz Behn is actively working to expand the reach of her findings. “Our current work is to apply a framework to our entire cohort of adolescent girls and use this to dig deeper in our understanding of PCOS in this dynamic setting,” she said. 

Lina Sorg is the managing editor of SIAM News.