Although the hallmark of type 1 diabetes is the autoimmune destruction of insulin producing β-cells, necessitating exogenous insulin, living with diabetes comes with a high psychological burden. This “diabetes distress” can include feelings of being overwhelmed, powerless and hopeless, along with fears of acute and long-term consequences. It includes the stress and frustration of daily management and conflicts with family and health care providers (1, 2). This can develop into a vicious cycle of increasing distress leading to emotional numbing and deteriorated glycemic control, which then in turn increases the psychological burden (3). Approximately 40–60% of individuals with type 1 diabetes experience moderate to high levels of this distress (4). This group is especially at risk for “diabetes burnout”; a state characterized by feeling exhausted, seen in combination with reduced self-care and diabetes management (2). These symptoms may be intense enough to fulfill diagnostic criteria for anxiety and depression. Individuals with type 1 diabetes are believed to have at least double the risk of various mental disorders and even suicide (5–7).

The situation is further complicated as about 50% of people with type 1 diabetes develop insulin resistance (8), worsening prognosis. To further complicate the picture; insulin resistance in the blood–brain barrier can occur without whole body insulin resistance, making it harder to detect, but with potential harm to the brain (9, 10). Alarmingly, there is an estimated loss of 100 days of life for each year a person’s HbA1c exceeds 58 mmol/mol (7.5%) (11).

Insulin resistance, and high levels of insulin and glucose are associated with mitochondrial dysfunction, neuroinflammation, neurotransmitter imbalances, glucose hypometabolism, and lower volume in parts of the brain (5, 9, 12, 13), all of which are also associated with mental disorders (10, 14–16).

Numerous factors complicate the estimation of appropriate insulin dosage. Two key factors are the variability in the effect of insulin administered subcutaneously (17), and miscalculations of the carbohydrate content in food, which may lead to a mismatch with the estimated insulin dosage (18). For example, assuming a combined error of 25%, the absolute value of the error will be lower with lower insulin doses. For instance, a 25% error at a “true” carbohydrate intake of 80 g will result in a miscalculation of 20 g in either direction. In contrast, a “true” carbohydrate intake of 10 g will result in a miscalculation of just 2.5 g in either direction. An insulin pump might compensate for the 2.5 g miscalculation, while a miscalculation of 20 g will significantly interfere with daily life.

Studies of various designs show that a low-carbohydrate diet (under about 100 g/day) lowers HbA1c, even approaching the prediabetic values of <48 mmol/mol (<6.5%), reduces glucose variability, and time spent in hypoglycemia without reducing quality of life (19–23). The ketogenic diet (about 20- 50 g/day of carbohydrates) appears even more beneficial, with HbA1c levels approaching non-diabetic values <42 mmol/mol (<6.0%) and normalize weigh (24–32). In addition, nutritional ketosis protects against the potentially detrimental effects of insulin resistance, high levels of insulin, and high/variable levels of glucose, thereby probably preventing and even treating symptoms of psychiatric disease and neurodegeneration (33). See overviews of the proposed mechanisms elsewhere, for example (34–38).

A key contributor to reduced mental health is the unpredictable nature of type 1 diabetes. A typical experience is that glucose levels tend to drop significantly during even light physical activity and exercise. This is likely due to heightened insulin mediated activation of the signaling pathway controlling GLUT4 glucose transporter activity (39).This effect may be amplified by (i) absence of the glucose sparing effect of nutritional ketosis (40, 41), and (ii) the slightly higher insulin levels by subcutaneous injected insulin compared to the release of endogenous insulin by the pancreas into the portal vein (17). Insulin levels can therefore often become high enough to cause uncomfortable drops during normal daily activities like house cleaning or short walks. This might increase diabetes distress and induce a fear of the very real threat associated with hypoglycemia. However, most cases of hypoglycemia can be easily corrected, often resulting in different threat assessments between healthcare personnel and their patients. Therefore, distinguishing between an anxiety and a rational fear of hypoglycemia has treatment implications, but may be challenging in clinical practice.

This case-report illustrates how introducing a KD mitigated intense diabetes distress and even symptoms of severe anxiety and depression.

4. Monitor ketones: Aim for blood ketones between 0.5- 3 mmol/L (42). Initially measure a few times a week, then two times or less a month may suffice. If ketone levels are 3 mmol/L or higher, consume about 10-30 g of carbohydrates and administer matched insulin. Test ketone levels if you get sick, unusually thirsty, headache, nausea, stomachache, or the like.

5. Regarding other medication: Notify relevant healthcare personnel of the change in diet. Do not combine with SGLT-2 inhibitor. Regularly assess the need to adjust blood pressure medication and psychopharmaceuticals.

The following months, the combination of her reduced insulin need and the insulin pump, her glucose levels significantly stabilized, with increased time in range [glucose values between 4 and 10 mmol/L (72–180 mg/dL)] from 3 to 51%, reducing her HbA1c from 83 to 58 mmol/mol (9.7 to 7.5%) (See Table 1).

Table 1. Information from continuous glucose monitor.

From 2 days after the emergency session and onwards, the clinician observed the patient’s amazement and joy of her reduced glucose variability, her relief of being believed in her motivation and efforts to manage her diabetes, and a clinically significant reduction in fear and depression.

Diabetic Ketoacidosis (DKA) is a potentially lethal condition marked by hyperglycemia [serum glucose >13.9 mmol/L (>250 mg/dL)], increased anion cap metabolic acidosis (anion gap >10–12, serum bicarbonate <18 mEq/L and/or pH <7.3), and ketosis (> 2 mmol/L) (43). Usually, DKA is seen in combination with insulin deficiency and/or increased amounts of counter-regulatory hormones like catecholamines, glucagon, cortisol, and growth hormone (44). The acidosis has been attributed to the increased level of ketones (43). However, the ketones acetoacetate and beta-hydroxybutyrate are not produced as acids but as conjugate bases, and acetone is neither an acid nor a base (45). The acidosis can therefore not be directly caused by the presence of ketone bodies. The acidosis is more likely a related consequence linked to the Krebs cycle’s maximum oxidation rate of Acetyl-CoA. Further lipolysis after the maximum oxidation is reached, will result in the release of a substantial number of protons per triglyceride molecule. And when these are not consumed by complex 1 in the electron transport chain, acidification may occur (45).

In contrast to DKA, in nutritional ketosis the blood pH remains withing normal limits and, glucose values are normal, but ketones are about 0.5–3 mmol/L (40, 46). Surprisingly, keto-adaptation has been observed to enhance the oxidation rate in ultra-athletes (47, 48). This could mean that the acidotic process associated with ketone production may be deferred through a KD, and possibly further postponed by including exercise (41, 46, 49).

Information is scarce on the incidence, prevalence, and risk factors of acidosis in type 1 diabetes patients in nutritional ketosis. One study reported a 1% incidence (24), compared to an estimated annual 5–8% incidence in the broader type 1 diabetes population (not on KD) (44). This information is relevant for this case, as it means that the patient is actually safer from a DKA on a KD.

In our opinion, a more pressing concern for type 1 diabetes patients is an increased risk of DKA during acute infections or pump malfunctions. However, if a pump with fast acting insulin malfunctions, the body’s insulin reserve depletes within hours. It is unclear how much ketones, induced by nutritional ketosis, will impact DKA development, as it is not the ketones themselves that are acidic. More research is needed. In addition, near normal glucose levels decreases tolerance for glucose fluctuations, thereby rapidly alerting the individual if insulin is needed due to hyperglycemia. And importantly, the improved predictability by combining KD with an insulin pump significantly reduces the mental burden related to diabetes, which was exactly what the current patient needed to reduce her anxiety and depression.

Euglycemic DKA involves acidosis with normal glucose levels and accounts for about 2.6–7% of known DKA cases (43). The main contributor appears to be the off-label use of SGLT-2 inhibitors. Other risk factors could include low-fat zero-carb diets, combined with prolonged fasting and intense exercise (50). Thus, the main contributor is not a KD, which is a high-fat low-carb diet that encourages the individual to eat when hungry and match insulin to maintain normal glucose levels. Keto-adapted people with type 1 diabetes often experience increased energy, satiety, and can thrive during intensive multi-day exercise, even when fasted (31, 32). As the current patient was not on SGLT-2 inhibitors but adhered to the KD aiming to better match her insulin injections to her need, the risk of euglycemic DKA was low.

Regardless of their diet and treatment, all individuals with type 1 diabetes must be vigilant about the risk of DKA.

4.2 Hypoglycemia and empty glycogen stores?

Intensive insulin treatment, leading to lower HbA1c levels, has been associated with a threefold increase in hypoglycemia frequency (51). Transitioning to a KD significantly reduces insulin needs, potentially causing a slight increase in hypoglycemic episodes, indicating that further insulin reductions are necessary. As the current patient suffered from severe hyperglycemia, the clinician calculated that the reduced insulin need due to KD would resolve the hyperglycemia without triggering anxiety, avoidance of insulin, nor lead to hypoglycemia. The insulin levels the patient managed to inject would be sufficient.

A primary concern other clinicians has is the belief that a KD depletes glycogen stores before ketone production occurs, potentially rendering glucagon injections ineffective during severe hypoglycemia. A small study (n = 10) found that glucagon injection resulted in a higher glucose rise after a high-carb week compared to a low-carb week. However, the rise was sufficient to rescue hypoglycemia in both groups (52), thereby rendering this finding irrelevant.

Three counterpoints to the glycogen storage total “depletion” notion include: (i) Similar levels of resting muscle glycogen stores and glycogen depletion after 180 min running were observed in keto-adapted endurance athletes and athletes consuming “high-carbohydrate” diet (49), potentially due to increased gluconeogenesis rates and a glucose-sparing effect of ketones (41, 46). (ii) During physical activity, muscles primarily use fat for fuel, and increased gluconeogenesis rates contribute to stabilize blood glucose (41). (iii) If hypoglycemia occurs, the brain uses ketones for fuel, mitigating the impact of reduced glucose availability in keto-adapted individuals (32, 53).

By informing the current patient of the benefits of becoming keto-adapted, including the additional fuel for the brain, the patient was able to accept the prospect of more insulin on board delivered by the upcoming pump treatment.

4.3 Cardiovascular disease (CVD)

CVD is the leading cause of death in type 1 diabetes patients (54). The main concern is that a KD, often relatively high in saturated fat, can increase low-density lipoprotein cholesterol (LDL-C) and therefore presumably the risk of CVD. However, there are five counter arguments: (i) Individuals with lower levels of LDL-C can become just as atherosclerotic as people with high LDL-C, if not more (55–57). (ii) A recent study found LDL-C not to be a significant risk factor in type 1 diabetes patients (58), suggesting that interventions should not be based solely on theory, but tested in the relevant populations. (iii) The understanding of CVD is evolving, now attributing more atherosclerotic properties to insulin resistance, inflammation and a subgroup of LDL particles, e.g., small dense LDLs (59, 60). (iv) A KD has been found to redistribute the fractions of small dense and big fluffy LDL, to lower circulating triacylglycerols, and to reduce other more important risk factors associated with high glucose levels and insulin resistance (59, 61). (v) The heart might benefit from running on ketones (62). Furthermore, patients with a normal BMI might be more likely to develop a lean-mass hyper-responder phenotype (LMHR) of very high LDL in combination with high HDL and low triacylglycerols (61), related to the carbohydrate restriction rather than the high saturated fat intake (63). This phenotype may not promote CVD risk in the same way as normal dyslipidemia, and there is therefore consensus that CVD risk assessment needs to be individualized according to metabolic state and other factors (64).The CVD risk was a not a current concern for this patient. In her state, her immediate priority was finding relief to take care of her children. Her main goal was to overcome her fear of insulin and resolve the toxic hyperglycemia. However, her HDL, triacylglycerol, and LDL levels did not indicate signs of LMHR, but her values will be closely monitored.

4.4 Strength and weaknesses

This report illustrates how a KD can be used in a normal clinical setting. A weakness is a lack of assessment scales for diabetes distress, anxiety, and depression pre and post the KD intervention, and a relatively short follow-up time.