Was she murdered by insulin injection?

She is profoundly hypoglycaemic. Like Ms CG in the exercise on Two diabetic patients in coma, this is probably the result of excessive insulin, either injected or the result of an insulin-secreting tumour of the pancreas (an insulinoma), since non-esterified fatty acids and ketone bodies are inappropriately low for someone who is so hypoglycaemic.

She was kept in hospital for several days, appeared to be well, and was given an oral glucose tolerance test (1 g of glucose /kg body weight). Plasma glucose was measured over 3 hours. The results are shown on the right.

She is obviously not diabetic - at all times after the test dose of glucose her plasma glucose is within the norma range (shown by the error bars on the control curve).

Since she was now well, she returned home. On a number of occasions over the next few weeks she again collapsed in a coma, and was treated by glucose infusion, when she recovered consciousness.

She was known to drink heavily (several large gin-and-tonics each evening, and sometimes more). Her husband said that often she drank her pre-dinner gin-and-tonics, but then did not eat her meal.

The small amount of sugar in the tonic water causes insulin secretion, then the alcohol inhibits gluconeogenesis and glycogenolysis, so prolonging the hypoglycaemic action of the insulin. (This is how a small drink before dinner stimulates appetite, by causing mild hypoglycaemia).

One night she again fell into a coma, and this time died shortly after arriving at the hospital. A blood sample was taken before she died, and again showed that she was profoundly hypoglycaemic, with low non-esterified fatty acids and no detectable ketone bodies. Her blood alcohol was 75 mg /100 mL - just below the legal limit for driving.

Her plasma insulin was also measured, and was found to be extremely high - 2000 mU /L.

Her hypoglycaemia seems to have been caused by an abnormally high plasma concentration of insulin. If this is endogenous insulin then she must have an insulinoma (an insulin secreting tumour of the pancreas). However, the glucose tolerance test carried out aft she had been in hospital for several days was normal. If she had an insulinoma then you would expect to see a very much lower plasma concentration of glucose both before and after the oral dose.

It is, of course, possible that this was injected insulin, either self-injected or injected by someone else in a (successful) attempt to murder her, in which case suspicion might fall on her husband.

What we need is some way of differentiating between injected insulin (which these days is recombinant human insulin - i.e. insulin made in micro-organisms using the human insulin gene) and insulin secreted by her pancreas.

This is possible, but it means that we have to go back to studies of a then novel way of measuring insulin developed in the mid 1960s.

He is obviously diabetic, with a high fasting blood concentration of glucose and a very much higher than normal response to the oral dose of glucose.

Samples were also taken for measurement of insulin at zero time and 1 hour after the glucose load.

At this time a new method of measuring insulin by radio-immunoassay was being developed in research laboratories of The Middlesex Hospital, and therefore both this and the conventional biological assay were used.

The biological method of measuring insulin is by its ability to stimulate the uptake and metabolism of glucose in muscle. Traditionally the diaphragm of a rat is used for this biological assay, since it is easy to dissect out, and can be cut into two halves of almost exactly the same weight. Both halves are incubated with [14C]glucose, one with and one without the sample containing insulin, in sealed "centre well" vials. At the end of the incubation an alkaline solution is injected into the outer compartment top trap carbon dioxide, and trichloroacetic acid is injected in to the inner well, containing the incubation mixture, to stop the reaction and drive off carbon dioxide. This assay is relatively tedious, and requires a rat for each sample to be analysed, as well as a number of rats to calibrate the assay against known amounts of insulin.

The then newly developed method of measuring insulin is by its ability to bind to anti-insulin antibodies, in competition with radioactively labelled insulin - this is radio-immunoassay, and is generally preferred because it is possible to assay a large number of samples at the same time. The antibody recognises, and binds to, the surface of the tertiary structure of the protein.

The radio-immunoassay was new, and had not yet been demonstrated to be as good as the biological assay. The assays were performed in the research laboratory as part of the process of validating the new assay method.

The results of the insulin assay were as follows (in mU of insulin /L):

This goes back to the early days of insulin, when it was not possible to measure it chemically, and its molecular mass was unknown. The only way of measuring insulin (until the development of radio-immunoassay) was by its biological activity.

The unit of insulin was originally that amount that would reduce the blood glucose concentration of a 1 kg rabbit by a given amount over 1 hour. Later work recalibrated the unit in terms of the effect on the metabolism of glucose by rat diaphragm incubated in vitro - originally measuring the volume of carbon dioxide released by manometry, then later by the release of [14C]carbon dioxide from [14C]glucose.

Insulin syringes are calibrated in units, not volume.

LC is obviously diabetic, in that he has a very low biological activity of insulin both in the fasting state and in response to a glucose load. However, there is something in his plasma that cross-reacts with the anti-insulin antiserum.

Obviously, this caused the researchers at The Middlesex Hospital considerable anguish. Their newly developed assay, which was quick, cheap and convenient, appeared to give a false negative result for this patient.

At this stage LC was treated with insulin injections, on the basis of the positive results of the glucose tolerance test and the biological insulin assay. Within a few weeks he was stabilised on a regime of insulin injections and dietary control, and has remained well since.

In an attempt to discover the cause of difference between the results of the biological assay and radio-immunoassay, the researchers performed gel exclusion chromatography on a pooled sample of normal serum, and determined insulin in the fractions eluted in each fraction by both radio-immunoassay and biological assay.

Gel exclusion chromatography

Proteins (and other macromolecules) can be separated according to their size by chromatography on columns of beads of gel that have small pores, so that smaller molecules spend more time within the pores of the support medium, and hence move through the column more slowly than larger molecules.

This is the technique of gel exclusion chromatography, also know as gel filtration or gel permeation chromatography.

By adding a series of coloured markers of known molecular mass to the sample undergoing chromatography, it is possible to calibrate the column, and hence estimate the molecular mass of the protein, by comparison of its elution volume with those of the standard markers.

In the columns used for this study, standard molecular mass markers were eluted as:

The insulin in each fraction eluted from the column was determined by biological assay (the graph on the left below) and radio-immunoassay (the graph on the right below):

The left-hand graph shows a peak of insulin biological activity around molecular mass 6000 (around fraction 23), which is about the molecular mass of insulin. There is also something that cross-reacts with anti-insulin antiserum in the same fraction, so we can be confident that this large peak represents insulin.

However, there is also something that cross-reacts with anti-insulin antiserum with a molecular mass of around 9000 (around fraction 10), but this has very low biological activity.

The researchers originally called this material that elutes around fraction 10 "big insulin", since it looks like insulin to radio-immunoassay, but has a molecular mass about 50% greater.

The next stage in their investigations was to treat each fraction eluted from the column and treat it with the proteolytic enzyme trypsin.

Again the insulin in each trypsin-treated fraction was determined by biological assay (the graph on the left below) and radio-immunoassay (the graph on the right below):

Trypsin treatment has had no effect on the immunologically active insulin - the right-hand graph is exactly the same s before trypsin treatment.

However, trypsin treatment has had an effect no the biological activity of "big insulin". It is now biologically active, and the curves for biologically active and immunologically active insulin are the same.

This suggests that trypsin has removed a part of the molecule of "big insulin", revealing a fragment that is biologically active insulin.

The obvious next step was to treat the serum sample with trypsin before gel chromatography, and again to measure insulin in each fraction by biological assay (the graph on the left below) and radio-immunoassay (the graph on the right below):

There is now only one peak of material that has both biological and immunological activity of insulin, with a molecular mass around 6000. The peak height for the radio-immunoassay is the sum of the heights of the two peaks seen previously. The peak height for the biological activity is the sum of the two peak heights seen after each fraction had been treated with trypsin.

This confirms that "big insulin" is indeed converted to insulin by treatment with trypsin.

When the researchers went back to look at the samples of blood they had taken from LC they found that almost all of the immunologically active insulin in his blood was "big insulin", which cross-reacts with anti-insulin antiserum, but has negligible biological activity.

More recently, the gene for human insulin has been cloned. Although insulin consists of two peptide chains, 21 and 30 amino acids long respectively, these are coded for by a single gene, which has a total of 330 base pairs between the initiator and stop codons. As you would expect for a secreted protein, there is a signal sequence coding for 24 amino acids at the 5’ end of the gene.

If the two chains of insulin (a total of 51 amino acids) are coded for by a single gene with 330 base pairs, then the product of translation of the insulin gene must be a peptide that is 110 amino acids long. allowing for a signal sequence coding for 24 amino acids leaves an additional 35 amino acids to be accounted for. These must be between the sequences of amino acids that will form the A and B chains of insulin.

This 110 amino acid product of translation of the insulin gene, shown on the right, is now known as pre-pro-insulin.

We now know that the sequence of events in insulin synthesis is as follows:

• Prepro-insulin is synthesised on ribosomes attached to the rough endoplasmic reticulum, and as it is being synthesised it is pulled into the endoplasmic reticulum by the signal sequence.
• Once synthesis is complete, the signal peptide is cleaved off, leaving an 86 amino acid peptide that is known as pro-insulin.
• The disulphide bridges that will hold the A and B chains together in insulin are then formed.
• In the Golgi apparatus the molecule is cleaved by a proteolytic enzyme (carboxypeptidase E) that has the same specificity as trypsin, leaving the A and B chains of insulin joined by disulphide bridges and a separate peptide that is known as the C peptide (so-called either because it connected the A and B chains in pro-insulin, or because if you have A and B chains then the next one logically is the C chain).
• Also in the Golgi apparatus, insulin and the C-peptide are packaged into vesicles, ready to be secreted in response to an increase in glucose entering the pancreatic beta-islet cells.

Obviously, LC was secreting pro-insulin rather than insulin. There are two possible reasons for this:

1. He lacked carboxypeptidase E, the enzyme that catalyses the cleavage of the C-peptide from pro-insulin
2. He had a mutation in his insulin gene so that it lacked one or both of the cleavage sites for carboxypeptidase E action.

You obviously cannot take a biopsy of his pancreatic beta-islet cells to test for carboxypeptidase E. However, you can test his circulating pro-insulin and see whether it is a substrate for either trypsin or carboxypeptidase E in vitro. If it is, then treatment with either enzyme would result in restoration of biological activity, as in the case of the "big insulin" fraction discussed above.

Alternatively, you could sequence his pro-insulin and see whether the amino acid sequence differs from normal. Or you could sequence his insulin gene from leukocytes or skin fibroblasts (the gene is present in all cells, even though it is only expressed in pancreatic beta-islet cells) and see whether the gene sequence is normal or not.

If his pro-insulin is normal, and can be cleaved to biologically active insulin by trypsin or carboxypeptidase E in vitro, then the fault must be with carboxypeptidase E, and not the insulin gene itself.

Radio-immunoassay is now the routine way of measuring insulin. How do you think the radio-immunoassay has been improved to overcome the problem that was encountered with LC in the 1960s?

Now that we know the problem we can make sure that the antisera used for radio-immunoassay react with insulin but not with pro-insulin. Equally, we can raise antibodies against the C peptide, permitting it to be measured as well.

The C-peptide is secreted together with insulin. It has no biological function, but clinically it is useful. Insulin is cleared by the liver within about 5 minutes of being secreted. This is important because there is a need to be able to vary the concentration of circulating insulin rapidly in order to maintain tight control over the plasma concentration of glucose, and terminate insulin action when it has exerted an adequate hypoglycaemic effect.

By contrast, C-peptide persists in the circulation for 15 - 30 minutes. There is no need to clear it rapidly, because it has no biological action.

If you want to measure the insulin response to a carbohydrate load, timing of taking blood samples is critical, because of the rapid clearance of insulin, and it is usual to measure C-peptide, as a surrogate marker for insulin secretion, because small errors in the timing of blood samples will have little effect on the results of C-peptide assay.

We can now come back to consider the death of Mrs PI.

The problem was to determine whether her abnormally high blood insulin, which led to her death from profound hypoglycaemia, was the result of over secretion of endogenous insulin as a result of an undetected insulinoma, or the result of deliberate injection of pharmaceutical insulin, as a means of committing either suicide or murder.

If the hypoglycaemia was the result of injection of pharmaceutical insulin, which is purified and does not contain C-peptide, then her blood concentration of C-peptide would be very low compared with the concentration of insulin.

There have been a number of murders and suspected murders associated with insulin, and many cases have been collected by Marks V and Richmond C in a book called Insulin Murders (RSM Press 2007).

Key points from this exercise:

• Excessive insulin, either as a result of an insulin secreting tumour or by injection can lead to profound hypoglycaemia.
• Consumption of alcohol with a small amount of carbohydrate can lead to hypoglycaemia. Carbohydrate stimulates insulin secretion and alcohol suppresses gluconeogenesis.
• Biological assay of insulin measures the stimulation of glucose metabolism in muscle tissue incubated with [14C]glucose in vitro; the rat diaphragm muscle is commonly used.
• Radio-immunoassay of insulin measures its ability to bind to anti-insulin antibodies, in competition with radioactively labelled insulin. This method is rapid and a large number of samples can be assayed at the same time.
• Insulin is measured in units of biological activity.
• Insulin consists of two peptide chains (A and B) joined by disulphide bridges. However it is coded for by a single gene, and is synthesised as a single, larger peptide (prepro-insulin) with a signal sequence that is cleaved in the rough endoplasmic reticulum. The resultant pro-insulin undergoes further post-synthetic modification in the Golgi apparatus, removing the C-peptide that connects the A and B chains.
• C-peptide is secreted together with insulin. It has no biological activity, but is useful in clinical chemistry because while insulin is cleared from the circulation within about 5 minutes, C-peptide survives for 15 - 30 minutes. Measurement of C-peptide gives a more reliable determination of the insulin response to aa test dose of carbohydrate.