It is refreshing to hear that some others out there are unhappy with the
state of play in HbA1c standardization.
 
Five years ago I decided that our cation exchange chromatography of HbA1c
needed a major rework. I set up a new assay using a Tosoh SP-NPR column. The
Tosoh column is a 2.5 micron nonporous resin with propyl-sulphonate
exchanger.

The small particle size and lack of pores gives very high efficiencies
compared to all other materials that I evaluated. I run a pH gradient rather
than the NaCl concentration gradient that is traditional. The eternal
problem in chromatography is chromatographic purity: that is, does the peak
represent one or more than one chemical species. The sharper the peaks and
the ability to resolve multiple peaks, in a typical sample, all increase the
probability that the peak of interest is homogenous. Typically we resolve 12
or more haemoglobin species in an 8-minute run.

When we first set this method up (5 years ago) and calculated %HbA1c based
on areas under the peaks we obtained values for HbA1c that were 20% less
than the values assigned to reference materials that had been checked
against DCCT materials. My assertion is that the HbA1c peak in our
high-resolution chromatography is more likely to be pure than the peak in
systems of lesser resolving power. If other haemoglobin species co-elute
with HbA1c then the HbA1c value will be overestimated.

We bought stocks of lyophilized calibrators 5 years ago and periodically
check these materials against samples distributed by the Australian Quality
Assurance Program (AQAP). I am lead to believe that the values assigned to
the AQAP materials are traceable to the DCCT calibrators. Based on these
observations we think that the DCCT values have been adjusted down and our
method now runs (un-adjusted) about 10% less than DCCT values. We routinely
use six-point calibration to increase our 'raw' estimates of HbA1c so that
the results we report are comparable to DCCT values.  

HbA1c is a different to most other analytes as it is a ratio rather than an
amount that is measured. Any reference method will involve physical
separation of the various haemoglobin species. It seems to me that any
discussion of the various methodologies should include quantitative measures
of the efficiency of this separation. Parameters such as plate counts and
the resolution factor 'R' (defined in all the standard textbooks of 'small
molecule' chromatography) are well known to specialist chromatographers.
There is a dearth of such discussion in most Biochemistry QC programs.
Collecting data, ad nauseam, on the performance of laboratories will achieve
little if the analytical methodology is flawed.

The manufacturers are under competitive pressure to provide analysers with
short run times. The fundamental principles of chromatography still apply:
resolution must be traded off against speed. Fast analysers necessarily
involve some compromise. 

High resolution chromatography allows detection of glycosylated forms of
haemoglobin variants such as glycosylated HbS. Given that HA1c is
exclusively formed by glycosylation of HbAo it is logical to report HbA1c as
a percentage of the HbAo present rather than as a percentage of all the
haemoglobins present in the sample. 

The ratio of HbA1c to HbAo will still reliably reflect the 'integrated'
blood glucose concentration when haemoglobin variants are present. I suggest
that the ratio of HbA1c to the total Hb is really a hangover from the old
batch isolation technique. The original batch method was the elution of a
fraction containing HbA1c from an ion exchange resin. The absorbance of this
fraction was then compared to the absorbance of a whole blood lysate treated
when treated with Drabkins solution. This latter ratio was also appropriate
for low resolution chromatographic where very few peaks were resolved and
two partly fused peaks were euphemistically referred to as the 'HbA1c' peak
and the 'HbAo' peak.

Does anyone want to buy into this ratio argument?

Phillip Paull
Clinical Biochemistry, St Vincent's Hospital Melbourne



%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%