The advent of halogenated pyrimidines (bromodeoxyuridine, BrdU; idoxuridine
, IdU) and antibodies to recognize them has opened new horizons for the mea
surement of proliferation in human tumours. These precursors of DNA can be
given to patients and a single biopsy can be taken to measure in a flow cyt
ometer both the fraction of labelled cells and their rate of movement throu
gh the S phase. From these two parameters the potential doubling time, T-PO
T, can be calculated. To measure both parameters simultaneously a compromis
e is made in the time of assessing the labelling index (LI). LI should idea
lly be assessed after a very short interval, e.g. 0.5-1 h, to avoid the con
taminating influence of any cells dividing between injection and biopsy. Ho
wever, an interval of 4-8 h is considered necessary to assess T-s from the
relative movement of cells through the S phase. Several techniques exist to
correct for cell division if the interval is long. The simplest correction
, which only corrects for the division of labelled cells, is most widely us
ed. Downward correction factors of at least 10% are commonly applied, reduc
ing the observed LI values. In this paper we illustrate graphically the dep
endence of the appropriate correction factor on various cell kinetic parame
ters. The duration of G2 is the most critical parameter for both the size a
nd direction of any correction factor. The G2 phase has previously been sho
wn to be about three times longer in human tumours than in rodents. If G2+M
is as long as 6 h, the main artefact of the intervals between injection an
d biopsy up to 7 h is that the observed LI is too low because of division o
f unlabelled G2 cells. A correction of up to 10% is needed but in an upward
direction. A nomogram of probable correction factors as a function of samp
ling interval is provided. We show from flow cytometric data that G2+M may
be shorter than 4 h for head and neck tumours. It is recommended that the c
orrection factor established by gating the flow histogram should always be
checked against this nomogram, or that no correction factor should be appli
ed. We have used this mathematical approach to re-evaluate two sets of publ
ished LI data for rectal and colorectal tumours. We show that the mathemati
cal correction of each data point leads to a 30% increase in the median val
ue, compared to the simple gating procedure. We question whether other of t
he published series of LI values gained with BrdU or IdU may also substanti
ally underestimate the true LI values, if a simple gating procedure has bee
n used in an attempt to reduce the impact of divided S phase cells.