HONK, HHS or DKA? Reaching the correct diabetes diagnosis

A 54-year-old man is unwell for two days, presenting to the GP with polyuria and polydipsia in the setting of an upper respiratory tract infection. He is not known to have a prior history of diabetes but he was known to have impaired fasting glucose (IFG) a year ago.

On evaluation on admission his blood sugar is 60mmol/l: 

• Do you think he has hyperosmolar non-ketotic hyperglycaemia (HONK), hyperosmolar hyperglycaemic syndrome (HHS) or diabetic ketoacidosis (DKA)? 

• How would you investigate?

• How are you going to manage him immediately and within the first 24 hours?

• How will you manage him in the next 24 hours?

• How do you anticipate you will manage him in the long-term? 

There is considerable confusion and debate about HONK and HHS and there is an overlap in both conditions in about 30% of cases. HONK tends to present with a shorter, less protracted course. The patient usually has hyperglycaemia, but without ketosis. He may have hyperosmolality and the presentation tends to be just prompter. 

HHS, however, probably represents the other end of that spectrum, occurring usually in the setting of a protracted course of hyperglycaemia, with subsequent resultant ketosis. Patients may have bicarbonate depletion and may indeed have mild ketoacidosis.

Both of these syndromes are manifest complications of patients with type 2 diabetes and may indeed reflect the primary presentation of the condition, or a deterioration of type 2 diabetes in the setting of precipitant factors. The dilemma, however, sometimes lies between differentiating diabetic ketoacidosis of type 1 diabetes in contradistinction to type 2 diabetes, versus this spectrum of HONK/HHS. 

Diabetic ketoacidosis (DKA) tends to have a very rapid presentation and while the blood sugar is elevated, it is usually elevated at around 30mmol/mol and certainly not higher than 45mmol/mol. It is defined by a PH of < 7.3 with an anion gap acidosis, and the patient is more likely to have urine ketonuria and serum ketonaemia. It is the classical presentation of patients with type 1 diabetes and is a typical presentation for patients with type 1 diabetes in whom there is a deterioration because of a particular precipitant. 

I often describe the typical precipitants of type 1 diabetes and DKA under the heading of the four Is, with ‘I’ referring to infection, incompliance, infarction and new diagnosis of type 1 diabetes (or IDDM, old nomenclature). 

In contrast, HONK and HHS tend to have a much more insidious course and, when they do present, the patient tends to have very elevated blood sugars that are typically much higher than the 30s (often > 45mmol/mol) and can be as high as 60-70mmol/l. 

Typically, the PH is normal or, in the case of HHS, may be borderline. There is not usually an anion gap and the patient may have urine ketonuria, but does not often have serum ketonaemia. This may be the presentation of a patient with newly presenting type 2 diabetes, or more classically the presentation in the setting of deterioration with type 2 diabetes. 

The precipitants for DKA are considered in the purview of the four Is and, really, precipitants for HONK or HHS can be attributed to similar causality, ie. the four Is of infection, infarction, incompliance and any major illness. 

The most important point is that one looks for a precipitant, and it is especially important not to forget infarction (be it from myocardial infarction or otherwise), and not to be misled/falsely reassured by a patient’s younger age. 

The hallmark features of DKA include hyperglycaemia, but typically less than 30mmol/mol, with concomitant significant glycosuria, ketonaemia, ketonuria and a positive anion gap. 

HONK by definition, however, is defined by much greater elevations of glucose (usually > 60mmol/mol) with glycosuria, and a normal anion gap. The hallmark of HONK is increased serum osmolality, frequently more than 330 (calculation for serum osmolality is 2 x serum + urea + glucose [normal range 275-295]). 

The perils of HHS are those of fluid deficiency and volume depletion. This is in contrast with DKA where the pivotal problem is frank insulinopaenia, and secondly fluid depletion.

This fluid deficit is usually in the order of 100-200ml/kg. The patient may appear to have low sodium, which is often a ‘pseudo-hyponatraemia’, and if one corrects the sodium for the elevation, you will often find that the corrected sodium is normal. Correction is on average 2.4mmol/l sodium for every increase in 5.5mmol/l of blood glucose above 5.5mmol.

Equally, the patient may appear to have high potassium (pseudo-hyperkalaemia). In reality, potassium cannot get into the cells (cellular hypokalaemia) and it must be recognised that as the volume and subsequent insulin depletion is corrected the potassium will move into the cells and ultimately this will result in a potassium deficit that in real terms is 3-5mEq/kg. 

The patient may have renal dysfunction due to excessive diuresis (due to hyperglycaemia) with subsequent volume depletion. The resultant hypoperfusion gives rise to renal dysfunction, and this is before one excludes/establishes other concomitant issues/comorbidities that may include cardiac dysfunction, cardiac failure or infarction.

Finally, profound hyperglycaemia may contribute to profound hyperlipidaemia, (typically hypertriglyceridaemia) and this may complicate the overall picture of pseudo-hyponatraemia.

Assessment

The assessment of the patient is imperative to the interpretation of the severity of the illness, and ultimately to the interpretation of the severity of the spectrum of HHS versus HONK. This assessment will also determine the rapidity with which one needs to correct the metabolic disruption and volume depletion.

It is important to assess whether the patient is alert or drowsy. Equally, one needs to determine if the patient is febrile (which may point to infection), hypotensive (which may be a sign of volume depletion or infarction) and tachycardic (which can reflect any of the above).

One also wants to assess if the patient is dry (secondary to profound volume depletion which may respond well to aggressive fluid resuscitation) and simply smell the patient’s breath, which may confirm ketosis/acidosis (acetone smelling breath). 

Dysarthria and focal neurological deficits are critical to exclude, because protracted hyperglycaemia with volume depletion can lead to effective ‘sludging’ with subsequent infarction and the picture of stroke.

Naturally, it is important to assess the cardiovascular system (to exclude cardiac failure or an acute cardiac infarction), the respiratory system (to exclude a chest infection) and also to determine if there are any respiratory components that may indicate a respiratory compensation for ketoacidosis. An abdominal exam is essential to exclude abdominal pathology (eg. ischaemia) and the most common site for infection that is often overlooked is the skin (especially the feet). 

The investigations of choice will include plasma glucose, serum electrolytes and in particular serum urea, creatinine, sodium, potassium, bicarbonate and chloride, and if indicated an arterial blood gas. You can also calculate the anion gap. 

It is important to assess urine and plasma ketones and plasma osmolality. I always look at an ECG (to exclude infarction) and an FBC that may direct one towards infection.

Amylase and lipase – do they matter?

Patients with DKA can have mild abdominal pain that is unrelated to pancreatitis, and they may have mild amylase elevations that are similarly unrelated to pancreatitis. 

Equally, type 2 diabetes patients may have amylase elevations that do not typically reflect pancreatitis and therefore are unlikely to be helpful.

Fluid management in HONK

The hallmark of HONK and HHS is fluid depletion first, and insulin depletion second, and therefore the immediate rank order of treatment of HHS/HONK is with fluid repletion first, followed by insulin.

As mentioned, the fluid deficit averages 100-200ml/kg, which in the context of a 100kg patient represents 10 litres in total. One corrects this deficit with isotonic saline (or half normal saline in the hypernatraemic patient). 

The rapidity of fluid resuscitation will be determined by the patient’s clinical picture, and indeed their shock status (which in turn is reflected by their mental status, pulse and blood pressure). 

The recommendation would be to fluid resuscitate with 10-15ml/kg per hour (about one litre per hour for the first four hours, and certainly even in the most shocked patient not exceeding 15ml/kg in the first four hours).

Insulin repletion in HONK – which is better, IV or subcutaneous?  

In HONK or HHS there is relative fluid depletion, and with expectant relative hypoprofusion, one has to assume that subcutaneous treatment may be haphazard or inadequate. I recommend that the insulin replacement in patients with HONK or HHS should be intravenous so that one is guaranteed repletion that is consistent and persistent. 

It is important to remember that, in conjunction with insulin management, one may expect that the patient becomes hypokalaemic, requiring potassium replacement. 

If indeed the potassium is less than 3.3mmol/l at the onset of treatment, the patient should be managed with fluids and potassium in the first instance, followed by intravenous insulin infusion (and I would recommend starting at 0.1 unit per kg per hour). 

The clinical picture of the patient determines the resolution of HHS. If the patient is mentally alert, with a blood glucose level < 15mmol/l , and all other parameters have returned to normal (hyperosmolality), then the patient may eat. 

Switching off the infusion 

• When do we switch it off?  

• Do we continue subcutaneous insulin?

• Do we switch to medications? 

When a patient presents with HHS or HONK, hyperglycaemia may be contributing to profound glucose toxicity and insulin resistance, such that even if the pancreas is functioning and does have potential secretory capacity, it may not be able to function in the setting of the acute illness.  

One can gauge this by the determination of the insulin infusion rates within the preceding 24 hours (and if indeed the patient has been on an IV infusion for 24 hours prior to that, it would be important to look at that too). 

One can assess whether the hourly insulin infusion rates are consistently high, which reflects insulin resistance. If the infusion rate has been constant, without any downward trend, you may decide to commit to that total 24-hour dose of insulin therapy as one switches off the infusion and transition to multiple daily injection (MDI).

If, however, there is a diminishing insulin requirement in the preceding few hours  (for example, going from 10 units per hour initially to one unit per hour) then one can assume that the toxic milieu has resolved and is improving, as reflected by reducing insulin requirements.

If the patient is on one unit/hour, then maximally the patient might only need 24 units of insulin in a 24-hour period. The decision can be made as to whether to convert to insulin MDI directly, or indeed to trial with medications, or revert to whatever the patient’s preceding drug regimen was. The HbA1c would also be helpful. If it has been normal prior to admission, one can feel relatively confident of returning the patient to their pre-admission medications. 

If, however, the HbA1c has been elevated and consistent with very poor control prior to admission, then one needs to realise that it may not be enough to return that patient to their pre-admission regimen.

Back to the case  

So, if we return to our patient to answer the questions posed above: 

• He was afebrile, with a pulse rate of 86, BP 120/80, respiratory rate 18/min and O2 Sats of 94%

• His BMI was 50. He was mentally alert. On respiratory exam he was noted to have a wheeze. His cardiac exam was normal and he did not have any focal neurological deficit

• His blood glucose was 56mmol/l

• His sodium was 123 (and corrected for glucose, he had a true sodium of 146)

• His potassium was 5.3, urea 13.7 with a creatinine of 164 (with a known baseline of 82) 

• His serum osmolality was 333.5 (2 x sodium + potassium + urea + glucose)

• His urine ketones were negative

• His FBC showed a Hb of 16.4 (compared to a Hb of 14) and he had leukocytosis with a WCC of 12.3

• His arterial blood gas showed a PH of 7.38 with bicarbonate level of 27

• His ECG was normal. 

Clearly, the impression is that this patient has progressed from IGT to frank diabetes (type 2 diabetes), and the precipitant here has been a chest infection. 

He has an obese phenotype to support the diagnosis of type 2 diabetes, as well as a preceding history of IGT. He has no other medical history. He is mentally alert and does not show any signs of hyperviscosity or stroke. He is not tachycardic or haemodynamically compromised. 

He has pseudo-hyponatraemia but with a corrected sodium of 146. He has serum hyperosmolality and the diagnosis would be consistent with HONK.

One would expect based on his weight (over 100kg) that he would be fluid depleted by at least 10 litres. However, he is mentally alert and I think it is reasonable to correct him at 10ml/kg/hour for the first four hours (one litre per hour for the first four hours) and thereafter convert him to three litres over 24 hours). He had potassium added to the first two of the seven litres. He needs (in addition to fluid resuscitation) insulin and the recommendation here is for IV insulin. 

He received a total of 124 units in that 24-hour period, and his hourly infusion rate started at 6.5 units per hour, titrated up to a maximum of 14.5 units per hour, and within the latter 12 hours brought down to five units per hour at the time of assessment 24 hours post-admission. 

Twenty-four hours post-admission the patient is well. He is mentally alert. He feels ready to eat. His blood sugar is 14 and his serum sodium is 135. His serum potassium is four and his serum creatinine is 120 (improved from 164). His bicarb is 28.   

• Do we switch off the insulin infusion? 

• Do we prescribe insulin? 

It was noted that his insulin requirements were quite significant (124 units in total in the past 24 hours) but that his requirements were significantly falling from 14.5 to five units per hour.

Nonetheless, in the setting of this resolution of infection and an inflammatory milieu and insulin resistance, he still requires five units per hour, which is on average 100 units in 24 hours. The decision was made to convert him to multiple daily injections. We titrated down by 25% giving him an average of 80 units. He was transferred to the ward. 

Will this patient be on insulin for the rest of his life?

This patient classically presents with newly diagnosed type 2 diabetes, in the setting of an infection and profound insulin resistance. There is an element of glucose toxicity and he needed huge amounts of IV insulin to resolve this (and the excessive dose also reflects insulin resistance). 

He had rapidly reduced titration of insulin, however, suggesting inherent pancreatic function and the expectation would be that indeed he would transition off insulin very quickly and be ready to be challenged with medication.

Maeve Durkan is a consultant in diabetes, endocrinology and metabolism at the Portiuncula Hospital and University Hospital Galway