Introduction

Dihydropyrimidine dehydrogenase (also known as DPD,
dihydrouracil dehydrogenase, dihydrothymine dehydrogenase, uracil reductase, EC
1.3,1.2) is the initial rate-limiting enzymatic step in the catabolism of not
only the naturally occurring pyrimidines, uracil and thymine, but also the
widely used antimetabolite cancer chemotherapy drug, 5-fluorouracil (5-FU).[1,2]
As shown in Figure 1, DPD occupies an important position in the overall
metabolism of 5-FU, converting over 85% of a standard dose of administered 5-FU
to dihydrofluorouracil (5-FUH₂), an inactive metabolite, in an enzymatic step
that is essentially irreversible.[3]

It is true that anabolism is clearly critical for 5-FU cytotoxic
action through conversion of 5-FU to the "active" nucleotides
5-fluorodeoxyuridine monophosphate (FdUMP), 5-fluorouridine triphosphate (FUTP),
and 5-fluorodeoxyuridine triphosphate (FdUTP). These important active
metabolites are, in turn, responsible for inhibition of cell replication through
inhibition of thymidylate synthase, or through incorporation into RNA or DNA,
respectively. Nevertheless, it is pyrimidine catabolism through the
rate-limiting step, DPD, that controls the availability of 5-FU for anabolism
and thus occupies a critical position in the overall metabolism of 5-FU.

Evidence of Importance of DPD Activity to 5-FU Pharmacology

DPD—as the rate-limiting step in pyrimidine catabolism—has
recently been shown to play a critical role in determining the clinical
pharmacology of 5-FU (Table 1). In particular, it has been demonstrated that DPD
accounts for much of the variability that has been noted in clinical studies
with 5-FU. This includes both intrapatient variability, as well as interpatient
variability.

DPD activity has been observed to follow a circadian pattern in
both animals and humans.[4-6] Studies in rats on a 12-hour light/12-hour dark
schedule have demonstrated that hepatic levels of DPD follow a pattern that can
be plotted on a cosine wave.[4] This pattern was completely reversed in another
group of rats on an inverted 12-hour light/12-hour dark schedule. In patients
receiving continuous-infusion 5-FU by automated pumps, sampling DPD activity in
peripheral blood mononuclear cells over a 24-hour period has also been shown to
exhibit circadian patterns when plotted on a cosine wave.[6] Serum samples
obtained at the same time from the same patients have been shown to have serum
5-FU concentrations that were also characterized by a circadian pattern, which
was essentially inverse to the DPD circadian pattern (Figure
2).

Data from this study suggested that perhaps DPD was responsible
for the circadian variation in 5-FU, leading some chemotherapists to propose
time-modified 5-FU infusions to optimize drug delivery during a 24-hour period.
Such regimens have been suggested by some oncologists, particularly those in
Western Europe, to have a potential benefit in the treatment of certain human
cancers.[7]

What Causes the Variation?

For the past 4 decades, it has been unclear as to why the
pharmacokinetics of 5-FU have been so variable, with half lives (t_1/2) ranging
from around 4 minutes to 25 minutes after an intravenous bolus (Table
2).[8]
Because of the critical position of DPD in the metabolic pathway, it was
hypothesized that the variation in pharmacokinetic characteristics might be
secondary to variability of DPD enzyme activity among different individuals.
Population studies were undertaken assessing DPD enzyme activity initially in
tissues (peripheral blood mononuclear cells and liver) from healthy individuals.
DPD was shown to vary from individual to individual, with a normal distribution
pattern (bell-shaped curve) showing a six-fold variation from the lowest to the
highest values (Figure 3).[9,10]

Essentially the same pattern of DPD activity has also been
observed in the peripheral blood mononuclear cells of both breast and colorectal
cancer patients, although, interestingly, the normal distribution is shifted to
the left, with lower median DPD activity in the breast cancer patient
population.[11] The mechanism for this latter observation is unknown. The wide
variation in DPD activity observed in the populations described above is thought
to account for the wide variation in the half-life (and pharmacokinetics)
observed in patients treated with 5-FU.[8]

Variation in DPD Activity

While most patients tolerate 5-FU reasonably well, over the past
4 decades, a number of patients have developed severe, at times
life-threatening, toxicity after standard doses of 5-FU.[12-15] Because these
patients demonstrated exaggerated normal toxic side effects (as if they had
received an overdose of drug), it was hypothesized that these patients were
deficient in a catabolic enzyme that resulted in more 5-FU being present over
time.

Initial studies in these affected patients demonstrated elevated
uracil and thymine levels, suggesting a deficiency in the first step in
pyrimidine catabolism, DPD.[12-14] Subsequent studies demonstrated that many of
these patients were indeed DPD deficient. It is now clear that an additional
small percentage (< 3%) of such patients have DPD activity that is
significantly below the normal distribution seen in the average person.[15]
These individuals are at significant risk if they develop cancer and are given
5-FU. This is a true pharmacogenetic syndrome, with symptoms being
unrecognizable until exposure to the drug.[15]

Variation in DPD activity has also recently been shown to be
responsible for the apparent variable bioavailability of 5-FU, which over the
years has led to the recommendation that 5-FU not be administered by the oral
route. An explanation for the erratic bioavailability of 5-FU has previously
been unclear, particularly since 5-FU is a small molecule with a pK_a that should
favor excellent absorption and bioavailability. Recent studies using DPD
inhibitors in animals have demonstrated that, following inhibition of DPD, the
pharmacokinetic pattern resulting from oral administration of 5-FU is
essentially the same as that produced by intravenous administration, resulting
in essentially 100% bioavailability.[16]

While it has long been well known that variability in the
activity of pyrimidine anabolic enzymes is important in determining the
antitumor effectiveness of 5-FU, much less attention has focused on the
variability in activity of the pyrimidine catabolic enzymes. Tumors that are
resistant to 5-FU have been shown to express increased levels of DPD
activity.[17] With the development of the quantitative polymerase chain reaction
(PCR) to measure DPD mRNA, increased expression of DPD mRNA has been
demonstrated in tumors of patients who were resistant to 5-FU.[18]

Developing a Pharmacologic Strategy Based on DPD

The variability in DPD levels in both normal and tumor tissues
provides a basis for the chemotherapist to consider either altering the dose of
the fluoropyrimidine drug or inhibiting DPD in order to eradicate the
variability in 5-FU pharmacology. The presence of increased tumor DPD could lead
to a decision to increase the dose of 5-FU (or a 5-FU prodrug) to overcome the
increased catabolism of 5-FU within the tumor, thus increasing the amount of
5-FU that could be converted to active anabolites (Figure
1).

An alternative approach is to use known inhibitors of DPD,
usually with a lower dose of 5-FU (or the use of a 5-FU prodrug; eg,
capecitabine [Xeloda]), to directly inhibit 5-FU degradation within the tumor,
thereby permitting the existing 5-FU (even if present in low concentrations) to
be converted to active anabolites. Inhibiting DPD in 5-FU-susceptible host
tissue, such as gastrointestinal mucosa and bone marrow, should also make dosing
from patient to patient less variable, the latter being accomplished through
lessening the variability in pharmacokinetics, bioavailability, and the
resultant host toxicity.