A hyperglycaemic hyperosmolar state in a young child: diagnostic insights from a quantitative analysis
From the 1Hospital das Clinicas, Faculty of Medicine of Ribeirão Preto, University of São Paulo, Ribeirão Preto, Brazil, 2Department of Critical Care Medicine, Hospital for Sick Children and Departments of Anaesthesia and Medicine, University of Toronto, and 3Division of Nephrology, St Michael's Hospital, University of Toronto, Toronto, Canada, and 4Division of Nephrology and Department of Medicine, Stellenbosch University, Cape Town, South Africa
Address correspondence to Professor M.L. Halperin, Professor of Medicine, University of Toronto, St Michael's Hospital Annex, Lab #1, Research Wing, 38 Shuter Street, Toronto Ontario, M5B 1A6, Canada. email: mitchell.halperin{at}utoronto.ca
 |
Summary |
|---|
 Top Summary IntroductionThe patientThe conundrumThe consultationAfter the adjournmentConcluding remarksAppendix 1: Estimating the...Appendix 2: Clues to...Appendix 3: Estimating the...AcknowledgementsReferences |
|---|
This teaching exercise demonstrates how the application of principles of physiology can identify the cause of a severe degree of hyperglycaemia (plasma glucose concentration 80 mmol/l) in a very young patient with newly diagnosed diabetes mellitus, determine whether the patient has diabetic ketoacidosis, and highlight the potential risks for this patient on admission and during initial therapy. A consultation with Professor McCance was sought to determine whether this patient had an unusual degree of ‘insulin resistance’. There were also uncertainties regarding the acid–base diagnosis. The patient did not appear to have an important degree of metabolic acidosis as judged from his pH of 7.39 and plasma bicarbonate concentration of 20 mmol/l in arterial blood; hence the diagnostic impression was that he had a hyperglycaemic hyperosmolar state. However, his plasma anion gap was significantly elevated, and remained so for 60 h, despite the administration of insulin. Issues in management concerning the basis for this severe degree of hyperglycaemia and how to minimize the risk of developing cerebral oedema are addressed. The missing links in this interesting story emerge during a discussion between the medical team and their mentor, Professor McCance.
|
Introduction |
|---|
|
TopSummaryIntroductionThe patientThe conundrumThe consultationAfter the adjournmentConcluding remarksAppendix 1: Estimating the...Appendix 2: Clues to...Appendix 3: Estimating the...AcknowledgementsReferences |
|---|
This clinical teaching exercise focuses on the approach to the patient who has a severe degree of hyperglycaemia, and the challenges in diagnosing and treating diabetic ketoacidosis (DKA). Once again, the hero of our clinical detective story is the imaginary consultant Professor McCance, who practiced medicine
70 years ago. Regular readers will know that his overall objectives are to demonstrate the application of principles of integrative physiology at the bedside together with a quantitative analysis to reach a more accurate clinical diagnosis, reveal the underlying pathophysiology, and help design better therapy. |
The patient |
|---|
|
TopSummaryIntroductionThe patientThe conundrumThe consultationAfter the adjournmentConcluding remarksAppendix 1: Estimating the...Appendix 2: Clues to...Appendix 3: Estimating the...AcknowledgementsReferences |
|---|
A 15-month-old, 12 kg male was brought to the emergency room with clinical and laboratory data consistent with a diagnosis of diabetes mellitus in very poor control. Polyuria and polydipsia had been very prominent over the past 3 days, and he had vomited on several occasions. In this interval, his intake was predominantly fruit juice. On more detailed questioning of his parents, the duration of his illness was >1 month, and characterized by poor appetite, failure to gain weight, and a progressive decline in physical activity. He was not known to have diabetes mellitus prior to this admission.
On physical examination, he was drowsy, but responded to external stimuli. Blood pressure was 110/52 mmHg, heart rate was 158 bpm, and respiratory rate was 30 breaths/min with deep respirations. His extremities were cool and capillary refill time was 6 s; the remainder of his physical examination was unremarkable. The major laboratory data obtained in the first 12 h of therapy are shown in Table 1. Of note, he had an unusually high plasma glucose concentration (PGlucose) of 80 mmol/l (1440 mg/dl).
|
After several unsuccessful attempts to establish an intravenous access, an intraosseous line was inserted. He was given a bolus of isotonic saline (10 ml/kg over 1 h) and an infusion of insulin (0.1 U/kg/h). A urinary catheter was inserted 3 h after admission to obtain reliable urine collections.
|
The conundrum |
|---|
|
TopSummaryIntroductionThe patientThe conundrumThe consultationAfter the adjournmentConcluding remarksAppendix 1: Estimating the...Appendix 2: Clues to...Appendix 3: Estimating the...AcknowledgementsReferences |
|---|
There were a number of issues that troubled the members of the medical team who were managing the case. They were uncertain why their patient had such a severe degree of hyperglycaemia, whether the correct diagnosis was a hyperglycaemic, hyperosmolar state or diabetic ketoacidosis (DKA), why his PGlucose did not fall at the expected rate after the administration of insulin, whether this implied an unusual degree of insulin resistance, and what the optimum strategy for therapy should be. With respect to the latter, two major opinions were offered. One was that the treatment implemented was ideal, because it had followed the recommendations of the consensus report on the therapy of DKA in children.1,2 The other opinion expressed doubts about these recommendations because of the risk of cerebral oedema, a dreadful complication of this condition that continues to occur with unacceptable frequency.3 To seek clarification, they decided to consult with Professor McCance, their kindly ‘guru’, presenting him with a short list of questions about the diagnosis and management of the patient to focus the discussion (Table 2). Professor McCance agreed to meet with the team at the patient's bedside the following morning, and asked the intern to try to find additional information about the urine flow rate in their patient and in other young patients with severe hyperglycaemia.
|
|
The consultation |
|---|
|
TopSummaryIntroductionThe patientThe conundrumThe consultationAfter the adjournmentConcluding remarksAppendix 1: Estimating the...Appendix 2: Clues to...Appendix 3: Estimating the...AcknowledgementsReferences |
|---|
Professor McCance could never resist a challenge and so, brimming with enthusiasm, he joined the team on their morning rounds. ‘The issues raised in this case are the same ones that troubled me when I practiced clinical medicine, and I see that the answers still elude the modern generation of physicians. I shall rely on my old stand-bys: quantitative analyses, principles of integrative physiology, and a broad examination of the problem to see whether we can come up with a better way to understand why your patient had these striking findings. Our first task will be to identify the major risks for your patient on admission. Next we must ensure that our therapy does not cause additional dangers for the patient. So: what are the major threats on admission to hospital for a young patient with hyperglycaemia?’ asked Professor McCance, looking at the intern who had admitted the young boy.
Question 1. What are the major threats on admission for a patient with a severe degree of hyperglycaemia?
The intern conceded that she had been quite alarmed by the very high PGlucose and ‘effective plasma osmolality’ (PEffective osm), but that the rest of the team had not viewed these as ‘immediate emergencies’. In fact, they were more concerned that a large fall in PEffective osm during therapy might put their patient at risk of developing cerebral oedema, the major cause of death in this setting.4,5 They were concerned about the haemodynamic state of the patient, but were unsure about how much saline should be given initially, as they had read that very aggressive initial therapy could also be a risk factor for the development of cerebral oedema.6 Professor McCance agreed with their assessment of the immediate threat, and suggested that they move onto diagnostic issues, but promised to return to the issues of therapy and the risk of cerebral oedema. He posed his second question. ‘Why does this patient have such a high PGlucose?’
Question 2. Why does this patient have such a high PGlucose?
Physiology principle 1
Concentration terms such as the PGlucose are made up of numerators (content of glucose in the extracellular fluid (ECF) compartment) and denominators (the ECF volume) (Figure 1). Each should be analysed separately, and one should start by assessing the ECF volume, as balance data for glucose cannot be known with certainty. Nevertheless, with this degree of elevation of the blood sugar, there always is a large quantity of glucose retained in the ECF compartment.
|
Return to the bedside
The physical examination does not provide quantitative information about the ECF volume. In a similar way, the usual laboratory tests, including the plasma concentrations of sodium (Na+) (PNa), urea (PUrea), uric acid (PUric acid), and hormone levels are helpful, but only in qualitative terms; similar conclusions could be drawn for all measurements in urine. ‘For these reasons‘, said Professor McCance ‘I would rely on the haematocrit to provide quantitative data on the plasma volume,7 as we did on a previous occasion with a patient who suffered from cholera.8’ If we examine this information (haematocrit adjusted for age and in the absence of diseases associated with a low haematocrit) in conjunction with expected changes in Starling forces, extrapolations can be made to assess changes in the ECF volume in quantitative terms. It is equally valid to use the concentrations of total proteins or albumin in plasma to provide this quantitative assessment, he said.
The haematocrit was 0.45 on admission (Table 1), higher than the expected mean haematocrit for this patient's age (0.37).9 Therefore the plasma volume (and by extrapolation, the ECF volume) was decreased by 30% after hyperglycaemia developed (see Appendix 1) and certainly contributed to the high PGlucose by virtue of the smaller denominator. However, despite this marked degree of ECF volume contraction, the patient also had a large excess of glucose (the numerator) in his ECF compartment (Table 3). Therefore Professor McCance asked, ‘Why was so much glucose retained in his ECF compartment?’
|
Question 3. Why was there such a high content of glucose in the ECF compartment?
Physiology principle 2
To have a high content of glucose in the ECF compartment, there must be a high input and/or diminished output of glucose. If the renal excretion of glucose is very low, the quantity of glucose added need not be as large.
Return to the bedside
‘To have such a high PGlucose, one would expect both a high input and a reduced output of glucose. Since we cannot know how much glucose the patient ingested, we can deduce how much was formed in the liver from glycogen and protein’, said Professor McCance, who then proceeded to summarize what he knew about glucose input, the glucose pool size, and glucose output in the light of the data available on their patient (Table 3).
Glucose input
When endogenous sources are considered, muscle glycogen cannot be converted directly to glucose because muscles lack the necessary enzyme, glucose-6-phosphatase. However, during a burst of exercise (e.g. a seizure), L-lactate can be formed in muscle and some will be released into the circulation, taken up by the liver and/or kidneys, and converted to glucose. There was no history to suggest that this mechanism was a factor in our patient.
Glycogen in liver (8 mmol/kg body weight) is converted to glucose in states with low net insulin bioactivity (low insulin, high glucagon and other counter-insulin hormones). This occurs very early in the correct hormonal setting, and it cannot occur twice! Glucose is also produced during protein catabolism.10 From a quantitative perspective, 1 mmol of glucose is produced along with close to 2 mmol of urea in this metabolic pathway.11 Since there was no rise in the PUrea in the first 7-h interval and very little urea was excreted due to the low initial urine flow rate, protein catabolism did not represent an appreciable input of glucose.
Over this period, the patient received 45 ml of Ringer's lactate (6 mmol of lactate), which could make only a very small contribution to the glucose pool size.
Another source of glucose is an exogenous one via the intake of large quantities of sweetened beverages (fruit juices) that some patients consume to quench their thirst. One litre of fruit juice contains approximately 750 mmol glucose equivalents; an amount 60-fold greater than our patient's normal glucose pool size (ECF volume of 2.4 l x PGlucose of 5 mmol/l = 12 mmol) and close to the quantity of glucose in 2 l of osmotic diuresis (which typically contains 350 mmol/l glucose12). ‘To gain insights into this process‘, said McCance, ‘let us examine the data to look at the rate of excretion of glucose, and also to look for unusual changes in the PGlucose.’
Excretion of glucose
‘From 7 to 12 h, I estimate that he excreted close to 100 mmol of glucose. Since the decrease in his glucose pool size was only 40 mmol, the need for mass balance tells me that another source of glucose input is needed, and I suspect that it was glucose absorbed from the intestinal tract.’ Even though the patient was no longer drinking solutions containing glucose after admission, there could have been a large reservoir of glucose in a dilated stomach,13,14 as we discussed in an earlier consultation.15
At this point, the excited intern caught the attention of Professor McCance. ‘I’m embarrassed to tell you that we had even better data to support your speculation, but we thought it was just a laboratory error‘, she said. ‘We noted that the PGlucose rose from 80 to 108 mmol/l about 2 h after therapy began. Since we could not understand how this could happen, we presumed it was a laboratory error! Had I paid attention to the PGlucose rise, the large ‘occult’ input of glucose would have been more obvious.’ The intern then asked: ‘Are there other signs to suspect that stomach emptying occurred?’
Question 4. Are there other signs to suspect that stomach emptying occurred?
Physiology principle 3
A large input of glucose in a patient who lacks the actions of insulin should result in a large rise in the PGlucose, because the glucose pool size is small (see Appendix 2 for clues to detect stomach emptying). If there is an appreciable glomerular filtration rate (GFR), the filtered load of glucose should be high enough to cause a large increase in the osmotic diuresis, which will become evident by a large rise in the urine flow rate (see Appendix 3 for an estimate of the GFR).
The above findings would be even more noticeable if they occurred after the PGlucose and the urine flow rate had declined. Figure 2 shows large, transient rises in the urine flow rate of another patient with a very high PGlucose and possible rapid stomach emptying.
|
Another sign that stomach emptying may have occurred is a change in the PEffective osm—this will rise if the stomach contains hyperosmolar glucose, but it could fall if the stomach contains hypo-osmolar fluid (e.g. if a patient switched intake from fruit juices to water). In fact, in an earlier consultation,15 we discussed a patient with DKA who switched her intake to water without sugar and had a sudden, large fall in her PGlucose and PEffective osm during the second 100 min of marked polyuria (10 ml/min). At that time, we speculated that this could impose an important risk for the development of cerebral oedema.6 Perhaps this could explain why close to 5% of patients who develop cerebral oedema do so before therapy begins.
The paediatric registrar then posed a question, ‘Since some of the glucose excreted in the urine is glucose that was retained in the stomach, does this mean that this urine output should not be replaced with intravenous saline?’
Question 5. Since some of the glucose excreted in the urine is glucose that was retained in the stomach, should this urine output not be replaced with intravenous saline?’
The question was fielded by the paediatric consultant, who commented that the ‘consensus view’ for the management of DKA stated that replacing the urine output should not be a goal of therapy.1,2 ‘I have not read any papers giving a valid physiological reason why this is the correct advice’ he continued, ‘but you have supplied a plausible explanation this morning.’ Professor McCance thanked him for his input, and stated that he would like to try and strengthen the argument even further using another physiological principle.
Physiology principle 4
Glucose is an ‘effective osmole’ because it is confined mainly to the ECF compartment, with low concentrations in most body cells—e.g. skeletal muscle cells.
Return to the bedside
The hyperglycaemia present on admission is helping to defend the ECF volume. Since oxidation of glucose will be minimal as long as fatty acid and ketoacid levels are elevated (6–8 h), the majority of glucose disappearance in this time interval will be via excretion in the urine. To prevent a further fall in the ECF volume, glucose that was retained in the ECF compartment should be replaced with Na+ and Cl–. In contrast, glucose in the urine that was newly added from endogenous glucose production or from glucose that was retained in the stomach, should not be replaced with Na+ and Cl–, because it would lead to expansion of the ECF volume and possibly increase the risk of developing cerebral oedema.6 Accordingly, one has to calculate how much extra glucose is present in the body—really in the ECF compartment. As illustrated in Table 3, the patient's ECF contained 136 mmol of glucose and the usual glucose content in his ECF compartment was 12 mmol. It would therefore be reasonable to replace the first 125 mmol of glucose excreted in the urine with Na+ and Cl–. This means that we should replace the first 300-ml of urine with 150-ml isotonic saline plus KCl, as a typical value for UGlucose in a glucose-induced osmotic diuresis is 350 mmol/l.12 Any glucose excreted in excess of this quantity represents new added glucose and should not be replaced with saline. Of course, the urinary losses of Na+ and Cl– should be replaced as isotonic saline to avoid negative balances for these ions and his ECF volume will need to be re-expanded.
The team was excited by these new insights. The irrepressible intern had one more question, ‘Can we use changes in the PGlucose to infer that insulin is acting?’, she asked.
Question 6. Can we use changes in the PGlucose to infer that insulin is acting?
Physiology principle 5
Glucose will only be oxidized if a fat-derived fuel is not available.
Return to the bedside
The PGlucose will fall initially by dilution when saline is infused and subsequently due to glucose loss in the urine—neither of these effects are due to insulin actions. The oxidation of glucose starts later, after insulin has caused a large fall in the concentrations of fat-derived metabolites (fatty acids and ketoacids) in plasma. The conversion of glucose to glycogen requires the induction of enzymes and the appropriate generation of signals; there will be a delay of hours before this pathway can remove an appreciable quantity of glucose. Hence an initial fall in the PGlucose does not indicate that insulin acted. Furthermore, because there may be an input of glucose due to stomach emptying after admission, PGlucose may fail to drop or may even rise despite the actions of insulin. Hence the answer to this question is no.
Professor McCance drew the discussion to a close by stating that the initial underlying pathophysiology was likely to be insufficient actions of insulin that led to a modest degree of hyperglycaemia (i.e. new onset of diabetes mellitus). Although this patient had a high intake of glucose, like many other children with diabetes mellitus, he also had a second problem, an ‘occult’, high rate of stomach emptying. Therefore his PGlucose rose markedly because the newly absorbed glucose from his intestinal tract exceeded the quantity of glucose that could be excreted or metabolized quickly. Thus, a modest degree of hyperglycaemia was converted to a state with a very high PGlucose. ‘My diagnosis is diabetes mellitus in poor control with an unusually high PGlucose due to rapid stomach emptying‘, said McCance. ‘Moreover, I cannot use the failure to lower the PGlucose rapidly during initial therapy as an indication of insulin resistance’. And so the session ended, with professor McCance promising to return the following morning to discuss the remaining questions, and in particular whether ketoacidosis was indeed present. An adjournment, he believed, would give them all some time to think about these issues in more depth. And with these words, he rushed off to his research laboratory.
|
After the adjournment |
|---|
|
TopSummaryIntroductionThe patientThe conundrumThe consultationAfter the adjournmentConcluding remarksAppendix 1: Estimating the...Appendix 2: Clues to...Appendix 3: Estimating the...AcknowledgementsReferences |
|---|
The intern volunteered to begin the discussion. She said she would use the same principle of mass balance that her mentor had emphasized for glucose, to decide whether their patient had a significant degree of ketoacidosis. ‘My first step will be to calculate the content of ketoacid anions in the ECF compartment by multiplying the concentration of ketoacid anions in plasma by the ECF volume’ she began confidently. She knew that the concentration of ketoacids in plasma was very low in the normal state. ‘Unfortunately, the concentration of ketoacid anions in plasma was not measured‘, she continued, ‘but I was wondering whether we could estimate their concentration from the PAnion gap?’
Question 7. Can the concentration of ketoacid anions be estimated from the PAnion gap?
Physiology principle 6
The concentration of new anions in plasma on admission can be deduced by subtracting the anionic valance of albumin in plasma from the PAnion gap.
Return to the bedside
The first time the concentration of albumin in plasma (PAlbumin) was measured was at 12 h (it was 34 g/l). ‘Therefore I would rely on changes in the haematocrit to determine how much the plasma volume changed in this period and therefore what the PAlbumin would have been on admission; I estimated that it would be 40 g/l’, prompted Professor McCance. ‘When the PAlbumin is 40 g/l, we assume that it has a net negative charge of 16 mEq/l (or 12 mEq/l if we also ignore the cationic charge of K+ in plasma).16’ The intern quickly performed the calculations: ‘Since the PAnion gap on admission was 25 mEq/l, I subtract 12 mEq/l for the negative charge on albumin, another 3 mmol/l for the raised PL-lactate, leaving us with a rise in the concentration of other new unmeasured anions of about 10 mEq/l.’ ‘So the content of ketoacid anions in the ECF on admission would be 10 mmol/lxECFV of 1.7 l, or about 17 mmol (Table 4), and therefore I conclude that our patient did indeed have an appreciable degree of ketoacidosis!’ she exclaimed victoriously.
|
The medical registrar then suggested: ‘Surely it should then be possible to make a similar calculation for the content of bicarbonate
in the ECF compartment to determine the severity of the acidosis?’
Question 8. Can the content of 
in the ECF compartment be calculated to assess the degree of acidosis?
Physiology principle 7
The content of 
in the ECF compartment is the product of its concentration and the ECF volume.
To determine if there is a deficit of , the calculation should be done using estimated normal values for concentration and ECF volume and comparing it to calculated ECF content of on admission. In addition, one should look for other possible sources for input and output of .
Return to the bedside
Since the normal value for this patient's ECF volume would be 2.4 l and 25 mmol/l for his PHCO3, the content of in his ECF would be 60 mmol before ketoacidosis developed. On admission, with an ECF volume 1.7 l and a PHCO3 of 20 mmol/l, the content of in his ECF compartment was only 34 mmol (Table 4), a deficit of 26 mmol of in his ECF compartment (60 minus 34 mmol). He had a history of vomiting which may have led to an additional input of ; therefore an even larger acid load would be needed to leave him with only 34 mmol of in his ECF compartment. There is another source of addition of that may not be obvious; that exiting the ICF of skeletal muscles because of the high venous PCO2, as suggested in reference 8.
There are three points worthy of emphasis. First, because his ECF volume was contracted, the PHCO3 appeared to be ‘higher than expected’. Second, the degree of ketoacidosis need to be larger than the 17 mmol of ketoacids that we determined in his ECF compartment on admission to explain the deficit of 26 mmol of plus the extra added to his ECF compartment. Third, if there were more H+ produced than ketoacid anions currently accounted for in his ECF compartment, this suggests that he excreted some of these ketoacid anions with a cation other than H+ or 
(Figure 3). ‘This is what I call an indirect loss of NaHCO3‘, explained Professor McCance. The team was impressed that simple calculations based on data available in the chart were sufficient to confirm that their patient did have DKA. However, there were still some loose ends to be tied up. ‘Why did the rise in the PAnion gap persist so long?’, asked the paediatric consultant.
|
Question 9. Why did the rise in the PAnion gap persist so long?
Physiology principle 8
The PAnion gap is the difference in the concentrations of unmeasured anions and unmeasured cations in plasma. When ketoacid anions are converted to
, or excreted with , there will be a decrease in the PAnion gap and an increase in PHCO3.
Return to the bedside
As shown in Figure 4, the PAnion gap remained high for 60 h, even though the ECF volume had been re-expanded. Therefore, new anions were added, but the PHCO3 did not fall; hence these anions were added with a cation other than H+. An obvious source of these anions was the infusion of Ringer's-lactate, which contains an equal mixture of L- and D-lactate. It is likely that these retained organic anions (25 mEq) were D-lactate (total infusion 1.7 l and concentration of D-lactate is 15 mmol/l), because D-lactate is metabolized more slowly than L-lactate.17 ‘Accordingly, I cannot use the absence of an early fall in the PAnion gap in this patient as a marker of insulin resistance‘, Professor McCance added. ‘If D-lactate were indeed involved, there should be a fall in the PAnion gap and a commensurate rise in the PHCO3 at a later time. Do you have any follow-up data on the PAnion gap in this patient?’
|
As shown in Figure 4, the PAnion gap fell to normal and the PHCO3 rose by a similar amount after 60 h in this patient. Since there was no change in the ECF volume at this time, we can use the change in the PHCO3 to assess the production of HCO3-. In fact, there was a 1:1 relationship between the fall in the PAnion gap and the rise in the PHCO3, and the PHCO3 rose to 30 mmol/l. Hence these data are compatible with an initial slow and later a faster rate of metabolism of D-lactate.
‘In the final few minutes, let me address one other issue’ said Professor McCance. ‘What should the goals of therapy be in patients with DKA?’
Question 10. What should the goals of therapy be in patients with DKA?
A major goal of therapy in a young patient with DKA is to prevent the development of cerebral oedema, the major cause of morbidity and mortality in this setting.3 This could be due to expansion of the ECF volume of the brain and/or swelling of brain cells. With regard to the latter, it is essential to prevent a large fall in the PEffective osm.6 Hence the tonicity of the intravenous fluids should be based on the need to achieve this goal.
Physiology principle 9
When considering the effect of an intravenous input on the tonicity of body fluids, compare the tonicity of the infusion to that of the plasma when the urine output is low, and to that of the urine when the urine volume is large. The calculations for the ‘effective’ osmolality in plasma and in urine in a patient with hyperglycemia are shown in equations 1 and 2, respectively.
|
| (1) |
|
| (2) |
Return to the bedside
In hours 0–7, the urine output was 340 ml. The PEffective osm rose to 399 at 7 h. Since the effective osmolality of the infusate in this period was 300 mOsm/l while that of the patient was 356 mOsm/l, the rise in the PEffective osm could not be due to this infusate. Moreover, since the UEffective osm was 450 mOsm/l (350 mmol/l glucose, 50-mmol UNa + UK, plus accompanying anions), this also could not be the cause of the rise in PEffective osm. Accordingly, the rise in the PEffective osm could be due to a shift of water into cells (e.g. a seizure18) or an input of hypertonic glucose. Since there were no seizures and the PGlucose rose by a similar degree to the rise in the PEffective osm, the rise in the latter was probably due to an input of hypertonic glucose. The source of this glucose was either new glucose production (i.e. from protein, which would be accompanied by a large appearance of urea) or the ‘occult’ absorption of glucose from the upper gastrointestinal tract. Since the rate of urea appearance was low, McCance suspected that the ingested sugar was being absorbed from the patient's intestinal tract.
The paediatric registrar added some supportive evidence: ‘The major intake in this patient was fruit juice, which contains close to 750 mmol/l of monosaccharide.12 Therefore it seems likely that the basis for the initial rise in the PEffective osm was an input of glucose that was present initially in his stomach.’
Professor McCance wished to tie up one more loose end. He was concerned about why the PNa failed to fall for the 60 h while this patient was in the intensive care unit (Figure 5) and what potential risks this implied for the patient. He asked, ‘Why did the PNa fail to fall for 60 h? And what was responsible for the later fall in the PNa?’
|
Question 11. What was the basis for the prolonged hypernatraemia, and what could cause a later fall in PNa?
Physiology principle 10
When vasopressin acts, the medullary collecting duct becomes permeable to water, and therefore the urine osmolality (Uosm) should be virtually equal to the osmolality of the medullary interstitial compartment. Since glucose is the major urine osmole during a glucose-induced osmotic diuresis and the Uosm is close to 700 mOsm/l (Table 5), the UNa + UK will be equal to half of the Uosm, minus the contribution of glucose, urea and NH4 to the Uosm.
|
Return to the bedside
During the glucose-induced osmotic diuresis in this patient, the average value for UNa + UK was 100 mmol/l. Therefore the urine was a hypotonic solution relative to plasma, and excretion of this urine cannot cause a fall in the PNa as long as glucose continues to be a major urine osmole.
Once the PGlucose fell below 15 mmol/l, the urine contained little glucose, and UNa + UK (plus their accompanying anions) accounted for the vast majority of the urine osmoles (Table 5). Of great significance for body tonicity, the sum of the UNa + UK was more than twice the PNa + PK. Hence the kidneys were now able to lower the PNa by ‘desalination’ of body fluids,19 as shown in Table 5.
The intern, who had been involved in taking care of the patient since his admission, thanked Professor McCance for drawing attention to the basis for the late decline in the PNa. She was now concerned about potential dangers: ‘If this were chronic hypernatraemia, a large, acute fall in the PNa from ‘desalination’ might cause catastrophic brain herniation. And this could be even more dramatic if the patient drank water in response to thirst or if we had elected to infuse hypotonic solutions to treat the hypernatraemia.’ Fortunately the team had not administered hypotonic saline and their patient had made a good recovery, without any obvious neurological abnormalities.
The paediatric consultant still appeared deep in thought. ‘Upon reflection, I have one last question about this patient’, he said. ‘Why might a patient have Kussmaul respirations when there is a very mild degree of acidemia?’
Question 12. Why might a patient have Kussmaul respirations when there is a very mild degree of acidemia?
Physiology principle 11
With marked stimulation of ventilation in the absence of hypoxia, suspect a high concentration of H+ in the respiratory centre.
Return to the bedside
There will be a rise in the concentration of H+ near the respiratory centre if the PCO2 is high and/or the PHCO3 is low in this region. As judged from the arterial blood values on admission (Table 1), this did not seem to be the cause of the Kussmaul respirations. ‘Accordingly, we must examine the evidence more closely‘, said Professor McCance. ‘The PCO2 that reflects the stimulus for ventilation is the capillary (and cerebrospinal fluid, CSF) PCO2 in the vicinity of the respiratory control centre, and not that in the arterial blood‘, he continued.
This capillary PCO2 can be appreciably higher than the arterial PCO2 if a large amount of oxygen is extracted from each litre of blood delivered to this region. Because there is little reason to believe that cerebral metabolism is stimulated in this patient who is drowsy and who responds only to stronger stimuli, there are two possible reasons for having a high capillary PCO2: uncoupling of oxidative phosphorylation (e.g. ingestion of acetylsalicylic acid) or a very low blood flow rate. ‘I favour the latter explanation‘, said McCance. In support of this view, he speculated that this patient had a very low GFR on admission (Appendix 3), which was likely due to the very contracted ECF volume. Both the very large dietary monosaccharide load and the very long duration of the illness (>1 month as noted in the medical history) contributed to the development of an unusually higher than expected glucose-induced osmotic diuresis, which usually contains 40–50 mmol of Na+ and Cl– per litre of urine.12
On the one hand, the degree of ECF volume contraction appeared to be only 30% when the haematocrit was used to make this estimate. This is probably not enough to overcome autoregulation of his cerebral blood flow. On the other hand, this calculation depended on the assumption that he arrived in hospital with a normal red blood cell mass and a haematocrit corrected for age. Nevertheless, since his nutrition was poor for 1 month, he may have had a lower red blood cell mass on admission. Similarly, because of this poor intake, his PAlbumin may have been lower than in normal subjects, and this again needs to be considered if the PAlbumin on presentation were to be used to make a quantitative assessment of the ECF volume.
In summary, the presence of Kussmaul respirations without an obvious cause as judged from the arterial blood pH, PCO2 and PHCO3 values provided a clue to suspect that he indeed had a sufficiently high concentration of H+ in the respiratory centre to stimulate ventilation appreciably. Another clue, which suggested marked ECF volume contraction, was the very low GFR (Appendix 3), which led to a very high PGlucose. Hence McCance was able to draw support for the speculation that he had a high capillary PCO2 in this area. This analysis also pointed out that it is important to measure the haematocrit and brachial venous PCO2 on admission and every 4 h through therapy to help estimate the degree of ECF volume contraction and thereby the need to adjust the rate of intravenous fluid administration to minimize one of the risk factors for developing cerebral oedema.
|
Concluding remarks |
|---|
|
TopSummaryIntroductionThe patientThe conundrumThe consultationAfter the adjournmentConcluding remarksAppendix 1: Estimating the...Appendix 2: Clues to...Appendix 3: Estimating the...AcknowledgementsReferences |
|---|
Having dealt with all the questions posed by the team in some detail (Table 2), Professor McCance summarized: ‘My diagnosis is diabetic ketoacidosis and a larger than usual exogenous input of glucose in a patient who could not metabolize or excrete this glucose rapidly; yes, it is a hyperglycaemic hyperosmolar state, but one due to rapid entry of glucose from the stomach’. He viewed the hyperglycaemic hyperosmolar state as a diagnostic finding, but not a specific metabolic or pathophysiological entity. He did not believe that there was undue insulin resistance, in view of the fall in the PK after insulin was administered (Table 1). In addition, he stressed that it was the quantitative estimate of the content of
and ketoacid anions in the ECF compartment that revealed that ketoacidosis was present, even though the PHCO3 was not very low. The reason for the latter was the contracted ECF volume and the added due to vomiting. Since this patient received Ringer's lactate, the slow metabolism of D-lactate caused the PAnion gap to remain high, so that this calculated value could not be used to assess whether insulin was acting.
With respect to therapy, the most important goal was to prevent cerebral oedema by avoiding a large fall in the PEffective osm. In patients who present with near-normal PNa on admission, the development of hypernatraemia is often needed when the PGlucose falls to prevent a fall in the PEffective osm.20 Once the PGlucose falls below the renal threshold of 10–15 mmol/l, the urine will contain little glucose, and UNa + UK may be greater than the PNa + PK, leading to a prompt fall in PNa by ‘desalination’. This may have catastrophic consequences, especially if hypernatraemia has been present for >48 h. The patient must therefore be observed carefully throughout his acute illness to recognize and prevent this particular threat, hypotonic fluids should not be administered. A synopsis of the principles of physiology is provided in Table 6.
|
|
Appendix 1: Estimating the ECF volume |
|---|
|
TopSummaryIntroductionThe patientThe conundrumThe consultationAfter the adjournmentConcluding remarksAppendix 1: Estimating the...Appendix 2: Clues to...Appendix 3: Estimating the...AcknowledgementsReferences |
|---|
When normal, a 15-month-old 12 kg child has a blood volume of 0.9 l (75 ml/kg). With a normal mean haematocrit of 0.37, his red blood cell (RBC) volume would be 0.33 l and plasma volume 0.57 l (haematocrit = RBC volume/total blood volume). In contrast, with a haematocrit of 0.45 and the same RBC volume, his blood volume would be 0.73 l. Thus, with the same 0.33 l RBC volume, his plasma volume would be 0.4 l, reduced by 30%. Ignoring changes in Starling forces for simplicity, the ECF volume should have declined by approximately 30% of its normal value: from 2.4 l (20% weight) to 1.7 l.
|
Appendix 2: Clues to suspect rapid stomach emptying |
|---|
|
TopSummaryIntroductionThe patientThe conundrumThe consultationAfter the adjournmentConcluding remarksAppendix 1: Estimating the...Appendix 2: Clues to...Appendix 3: Estimating the...AcknowledgementsReferences |
|---|
When there is a large intake of fruit juice, for this glucose to enter the ECF compartment, the stomach must propel its contents into the intestinal tract where glucose will be absorbed. Supporting evidence for this hypothesis includes the initial 28 mmol/l rise in PGlucose from 80 to 108 mmol/l, the failure of PGlucose to fall appreciably over the first 7 h of therapy (Table 1), and the estimated glucosuria in this patient. Two other clues to suspect a bolus ‘infusion’ from stomach emptying are a large rise in the urine flow rate (Figure 2) and a change in the PEffective osm. If the patient drinks fruit juice or sweetened soft drinks, PEffective osm will rise. The converse is also true; if the patient drank pure water, a sudden, large fall in the PEffective osm could occur, which could increase the risk of developing cerebral oedema. In fact, close to 5% of patients who develop CE do so before therapy begins.4,5
|
Appendix 3: Estimating the GFR using glucose as the indicator |
|---|
|
TopSummaryIntroductionThe patientThe conundrumThe consultationAfter the adjournmentConcluding remarksAppendix 1: Estimating the...Appendix 2: Clues to...Appendix 3: Estimating the...AcknowledgementsReferences |
|---|
The GFR falls due to the deficit of Na+ created by the glucose-induced osmotic diuresis, which leads to ECF volume contraction, a lower blood pressure—hence the glomerular capillary hydrostatic pressure is decreased and the glomerular capillary colloid osmotic pressure is higher.
‘This calculation will be an ‘educated’ guess, as I am relying on estimates rather than hard data', said Professor McCance. The filtered load of glucose is equal to the PGlucose minus 10 mmol/l (renal threshold) multiplied by the GFR. Since the quantity of glucose excreted was 100 mmol in 5 h and the PGlucose was on average 60 mmol/l (Table 1), the GFR is only 2 l/5 h (100 mmol/50 mmol/l = GFR) or 0.4 l/h. The patient's normal estimated GFR/h for his age and weight is 2 l/h (his body surface area is 0.5 m2, so 120 ml/min/1.73 m2 is equivalent to 7.2 l/h/1.73 m2 and 2 l/h/0.5 m2). Hence he has a very low GFR, and his low rate of excretion of glucose is a major contribution to his hyperglycaemia. As shown in Table 7, a large amount of glucose will be excreted when there is both a significant degree of hyperglycaemia and a near-normal GFR. To maintain a steady state in this setting, an extremely large intake of glucose would be needed. In contrast, much less glucose will be excreted with a given rise in PGlucose when the GFR is reduced.
|
|
Acknowledgements |
|---|
|
TopSummaryIntroductionThe patientThe conundrumThe consultationAfter the adjournmentConcluding remarksAppendix 1: Estimating the...Appendix 2: Clues to...Appendix 3: Estimating the...AcknowledgementsReferences |
|---|
The study was supported by a grant from CAPES (#1309/06-4).
|
References |
|---|
|
TopSummaryIntroductionThe patientThe conundrumThe consultationAfter the adjournmentConcluding remarksAppendix 1: Estimating the...Appendix 2: Clues to...Appendix 3: Estimating the...AcknowledgementsReferences |
|---|
1. Dunger DB, Sperling MA, Acerini CL, et al. (2004) ESPE/LWPES Consensus statement on diabetic ketoacidosis in children and adolescents. Arch Dis Child 89 188–94.
2. Dunger D, Sperling M, Acerini C, et al. (2004) European Society for Paediatric Endocrinology/Lawson Wilkins Pediatric Endocrine Society Consensus Statement on Diabetic Ketoacidosis in Children and Adolescents. Pediatrics 113 e133–40.
3. Sperling MA. (2006) Cerebral edema in diabetic ketoacidosis; an underestimated complication? Pediatric Diabetes 7 73–4.[CrossRef][Web of Science][Medline]
4. Edge JA, Hawkins MM, Winter DI, Dunger DB. (2001) The risk and outcome of cerebral oedema developing during diabetic ketoacidosis. Arch Dis Child 85 16–22.
5. Glaser NS, Wootton-Gorges SL, Marcin JP, Buonocore MH, DiCarlo J, Neely EK, Barnes P, Bottomly J, Kuppermann N. (2004) Mechanism of Cerebral Edema in Children with Diabetic Ketoacidosis. J Pediatr 145 164–71.[CrossRef][Web of Science][Medline]
6. Carlotti A, Bohn D, Halperin M:. (2003) Importance of timing of risk factors for cerebral oedema during therapy for diabetic ketoacidosis. Arch Dis Child 88 170–3.
7. Love AHG and Phillips RA. (1969) Measurement of dehydration in cholera. J Inf Disease 119 39–42.
8. Zalunardo N, Lemaire M, Davids MR, Halperin ML. (2004) Acidosis in a patient with cholera: A need to redefine concepts. Q J Med 97 681–96.[Web of Science]
9. Oski FA, Brugnara C, Nathan DG. (2003) Diagnostic approach to the anemic patient. In Nathan DG and Oski FA (Eds.). Hematology of infancy and childhood 6th Philadelphia WB Saunders.
10. Jungas RL, Halperin ML, Brosnan JT. (1992) Lessons learnt from a quantitative analysis of amino acid oxidation and related gluconeogenesis in man. Physiol Rev 72 419–48.
11. Halperin ML and Rolleston FS. (1993) Clinical Detective Stories: A Problem-Based Approach to Clinical Cases in Energy and Acid-Base Metabolism 1st London Portland Press.
12. Halperin ML, Goguen JM, Scheich AM, Kamel KS. (1993) Clinical consequences of hyperglycemia and its correction. In Seldin DW and Giebisch G (Eds.). Clinical Disturbances of Water MetabolismNew York Raven Press pp. 249–72.
13. Schvarcz E, Palmer M, Aman J, Horowitz M, Stridsberg M, Berne C. (1997) Physiological hyperglycemia slows gastric emptying in normal subjects and patients with insulin-dependent diabetes mellitus. Gastroenterology 113 60–6.[CrossRef][Web of Science][Medline]
14. Fraser RJ, Horowitz M, Maddox AF, Harding PE, Chatterton BE, Dent J. (1990) Hyperglycaemia slows gastric emptying in type I (insulin-dependent) diabetes mellitus. Diabetologia 33 675–80.[CrossRef][Web of Science][Medline]
15. Davids MR, Edoute Y, Stock S, Halperin ML. (2002) Severe degree of hyperglycemia: novel insights revealed by the use of simple principles of integrative physiology. Q J Med 95 113–24.
16. Kamel KS, Cheema-Dhadli S, Halperin FA, Vasudevan S, Halperin ML. (1996) Anion gap: Do the anions restricted to the intravascular space have modifications in their valence? Nephron 73 382–9.[Web of Science][Medline]
17. Oh MS, Uribarri J, Alveranga D, Lazar I, Bazilinski N, Carroll HJ. (1985) Metabolic utilization and renal handling of D-Lactate in men. Metabolism 34 621–5.[CrossRef][Web of Science][Medline]
18. Welt LG, Orloss J, Kydd DM, Oltman JE. (1950) An example of cellular hyperosmolarity. J Clin Invest 29 935–9.[Web of Science][Medline]
19. Steele A, Gowrishankar M, Abrahmson S, Mazer D, Feldman R, Halperin ML. (1997) Postoperative hyponatremia despite isotonic saline infusion: A phenomenon of ‘desalination’. Ann Intern Med 126 20–5.
20. Hoorn EJ, Carlotti APCP, Costa LA, MacMahon B, Bohn G, Zietse R, Halperin ML, Bohn D. (2007) Preventing a drop in effective plasma osmolality to minimize the likelihood of cerebral edema during treatment of children with diabetic ketoacidosis. J Pediatrics 150 (in press).
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||