Population Study of Dihydropyrimidine

Dehydrogenase in Cancer Patients

By M.C. Etienne, J.L. Lagrange, O. Dassonville, R. Fleming, A. Thyss, N. Ren6e, M. Schneider, F. Demard,
and G. Milano

Purpose: We conducted a prospective study on a DPD activity, but that DPD activity was, on average, 15%
large set of cancer patients in an attempt to evaluate lower in women (0.194 nmol/min/mg protein) than in
the incidence of complete or partial dihydropyrimidine men (0.228 nmol/min/mg protein) (P = .03). No differ-
dehydrogenase (DPD) deficiency as found in peripheral ence was demonstrated between premenopausal and
mononuclear cells (PMNC). postmenopausal women. In patients treated with FU, the
Patients and Methods: One hundred eighty-five unse- risk of developing side effects was not linked to pretreat-
lected consecutive cancer patients were included. The ment DPD activity. FU-related toxicity was linked to FU
population consisted of 152 men (mean age, 62.1 years; systemic exposure. The correlation between pretreat-
range, 35 to 90) and 33 women (mean age, 59.2 years; ment DPD activity and FU systemic clearance (CI) was
range, 36 to 77). Sixty-eight were head and neck pa- weak (n = 90, linear regression r = .31, P = .002). Pre-
tients treated by a 5-day continuous infusion of fluoro- treatment DPD activity in patients who required a dose
uracil (FU; starting dose, I g/m 2 /d, with dose adaptation reduction was not significantly different from DPD activ-
based on pharmacokinetics) for which DPD activity was ity in patients who did not require dose modification.
measured 2 to 3 days before FU administration (94 cycles Conclusion: From the present study, it appears that
analyzed). PMNC-DPD activity was measured by a radio- total DPD deficiency is a rare event. Although pretreat-
enzymatic assay using carbon-14-FU. ment DPD activity cannot be a useful indicator for im-
Results: DPD activity in the entire population showed proving FU dose adaptation strategy, the identification
a unimodal distribution, which globally fits a gaussian of severe DPD deficiency (< 0.100 nmol/min/mg protein)
distribution. Mean and median DPD activity values were could lead to starting the treatment with a markedly re-
0.222 and 0.211 nmol/min/mg protein, respectively duced FU dose or even to using an alternative chemother-
(range, 0.065 to 0.559). No total DPD deficiency was apy regimen.
found. Multifactor analysis of variance showed that liver J Clin Oncol 12:2248-2253. © 1994 by American So-
function (biologic evaluation) and age did not influence ciety of Clinical Oncology.

F LUOROURACIL (FU) is commonly used in the

treatment of digestive, breast, and head and neck
have reported a significant link between FU pharmacoki-
netics and FU-related toxicity.6,8-10 We previously estab-
cancer. Recent advances in both basic and clinical re- lished that for continuous infusion, the FU-related toxicity
search have improved FU therapeutic efficacy through can be significantly decreased by adjusting FU doses
biochemical modulation using synergistic drug associa- based on midcycle plasma area under the curve (AUC)
tions (eg, folinic acid, dipyridamole, cisplatin, and inter- values.' It would be even more interesting to identify,
feron).' Consequently, prescription of FU chemotherapy before starting FU treatment, patients at risk for impaired
is currently increasing. FU treatments carry a nonnegligi-
FU Cl and subsequent toxicity.
ble risk of severe side effects2 3 and several attempts have
Dihydropyrimidine dehydrogenase (DPD) is the initial
been made to improve the FU therapeutic index, including
new continuous infusion schedules.4 Whatever the sched- enzyme in the catabolism of FU. " The clinical importance
ule, FU systemic clearance (Cl) exhibits wide interpatient of DPD has recently been demonstrated with the identifi-
variability.5 6' We recently evaluated the link between FU cation of severe toxicity in patients with suspected or
C1 and different physiologic parameters. None of the fol- proven DPD deficiency as found in peripheral mononu-
lowing factors were found to influence FU Cl: FU dose, clear cells (PMNC).12 '16We17 and others' 8 have demon-
hepatic function, nutritional status, or age.5 '7 However, FU strated that DPD activity in human PMNC was correlated
systemic clearance was demonstrated to be significantly to FU Cl. The next step for DPD investigation was to
lower in women as compared with men.5 Several studies conduct a prospective study to evaluate DPD activity in
PMNC on a large set of cancer patients to attempt to
evaluate the incidence of complete or partial DPD defi-
From the Centre Antoine Lacassagne, Nice, France; and Cancer ciency. We thus report a prospective study performed on
Center of Wake Forest University, Winston-Salem, NC. a population of 185 unselected cancer patients.
Submitted February 22, 1994; accepted June 24, 1994.
Supported by the FiddrationNationaledes Centres de Lutte Con- PATIENTS AND METHODS
tre le Cancer, Paris,France.
Address reprint requests to G. Milano, PhD, Centre Antoine La-
Materials
cassagne, 36 voie Romaine, 06050 Nice Cedex 1, France. All chemicals were obtained from Sigma Chemical Co (St Quentin
© 1994 by American Society of Clinical Oncology. Fallavier, France) and were of the highest purity available. Dihy-
0732-183X/94/1211-0004$3.00/0 drofluorouracil (FUH2) was kindly provided by Roche Laboratories

2248 Journal of Clinical Oncology, Vol 12, No 11 (November), 1994: pp 2248-2253

DIHYDROPYRIMIDINE DEHYDROGENASE IN CANCER PATIENTS 2249
8
(Neuilly/Seine, France). Carbon-14-labeled FU (55 mCi/mmol) was DPD activity according to a previously reported method.' The assay
from Amersham (Amersham, United Kingdom). Histopaque (Ficoll consisted of the incubation of 50 LL of cytosol (0.05 to 0.10 mg
and sodium diatrizoate) was purchased from Sigma Chemical Co. of cytosolic protein) with '4 C-FU (20 umol/L final), reduced 8/-
RPMI 1640 cell culture medium was from Whittaker Bioproducts nicotinamide-adenine dinucleotide phosphate (250 gmol/L final),
(Fontenay sous Bois, France). and magnesium chloride (2.5 mmol/L final). Total volume was 125
AL (in 35 mmol/L sodium phosphate buffer pH 7.5 containing sodi-
0
Patients umazide). The duration of incubation was 30 minutes at 37 C. The
reaction was stopped by addition of 125 ML of ice-cold ethanol
This prospective study was conducted on 185 consecutive and
followed by a 30-minute storage at -20"C. The samples were centri-
unselected cancer patients. Cancer localizations were as follows: 132
fuged (5 minutes at 400 x g) to remove proteins and the supernatant
head and neck, 38 digestive tract, six lung, three kidney, one breast, 4
was analyzed for the presence of 1 C-FUH2 using a previously re-
two chronic lymphocytic leukemia, two non-Hodgkin's lymphoma,
ported high-performance liquid chromatographic method.'9 Detec-
and one myeloma. The investigated population consisted of 152 men
tion was performed using a radioactive flow monitor (LD 506;
(mean age, 62.1 years; range, 35 to 90) and 33 women (mean age, 20
Berthold, Wildbad, Germany). As noted by other investigators us-
59.2 years; range, 36 to 77; 26 postmenopausal). The age distribution
ing crude cytosolic extracts of PMNC to determine DPD activity,
was not significantly different between men and women (Student's
the only FU catabolite observed was FUH2. Cytosolic protein con-
t test, P = .16). For 182 patients, hepatic function tests were per-
centrations were determined by the dye-binding method (Bio-Rad
formed in the month before DPD measurement (for patients with
SA, Ivry sur Seine, France) using bovine y-globulin as standard.
more than one DPD determination, liver function was checked before
Thus, enzyme activity was expressed as nanomoles of FUH2 formed
the first DPD measurement). Investigated parameters were y-gluta- per minute and per milligram of protein. Each sample was assayed
myl transferase (normal range, 8 to 76 IU/L), alkaline phosphatase
in duplicate. The sensitivity limit was 0.010 nmol/min/mg protein.
(normal range, 38 to 126 IU/L), total bilirubin (normal value, < 13
The interassay reproducibility (aliquots of a pooled PMNC suspen-
mg/L), and albumin (normal range, 32 to 45 g/L); one or more
sion) gave a coefficient of variation of 12%.
deviations from the normal values were observed in 58 patients. For
169 patients, DPD activity was measured before they received any
chemotherapy or at least 3 weeks after the last chemotherapy course. FU PharmacokineticInvestigations
The remaining 16 patients received chemotherapy that contained FU For the 68 patients who received FU chemotherapy, FU pharmaco-
1 to 3 weeks before DPD measurements. kinetic analysis was performed for the 94 cycles investigated for
Sixty-eight patients (58 men and 10 women) had head and neck DPD. The 5-day infusion of FU started at 8 AM on day 2. Blood
cancer treated by neoadjuvant chemotherapy with FU. Treatment sampling times were 5 PM on day 2, 8 AM and 5 PM on days 3
consisted of cisplatin (100 mg/m2 intravenously [IV] on day 1) fol- and 4, and 5 PM on days 5 and 6. Venous blood (5 mL) was drawn
lowed by a 5-day continuous infusion of FU administered with a in edathamil tubes and immediately centrifuged (10 minutes at 2,500
2
controlled-flow pump (starting dose, 1 g/m /d with dose adaptation at rpm) in the hospitalization ward. Plasma was transmitted to the
9 0
midcycle based on pharmacokinetics). Cycles were repeated every 3 laboratory and stored at -20 C until analyzed. FU concentrations
weeks (three cycles planned). Blood samples for DPD measurement in plasma were analyzed by high-performance liquid chromatogra-
were collected from these 68 patients 2 to 3 days before FU adminis- phy.21 The limit of sensitivity was 10 ng/mL. The interassay repro-
tration. Ninety-four cycles were studied in total: 20 DPD measure- ducibility (aliquots of a pooled plasma spiked with FU) gave a
ments before cycle 1, 37 before cycle 2, and 37 before cycle 3. For coefficient of variation of 6%. On the morning of day 4, the AUC
analysis of DPD distribution, the first DPD measurement was taken (AUCO-48h) was calculated to modify, if necessary, the FU dose given
into consideration. The FU-related toxicity was evaluated according in the second half of the cycle.' A dose reduction was thus performed
to World Health Organization criteria. in 22 cycles and a dose augmentation in three cycles. The mean
dose administered over the 5 days was 769 mg/m 2/d (range, 448 to
Collection of PMNC 1,156 mg/m 2 /d). For 90 of 94 cycles (four cycles had incomplete
Because of the previously reported circadian variability of DPD blood sampling), we also calculated the AUC over the duration of
8
activity in human PMNC," blood samples were collected between pharmacokinetic follow-up (AUC-105sh) and the FU Cl by dividing
8 AM and 11 AM to minimize the influence of circadian variability. the total dose administered over 105 hours by the AUCO-105h. AUCs
Approximately 20 mL of blood was collected in heparinized tubes, were calculated according to the trapezoidal rule.
transferred into a 50-mL tube, and mixed with 15 mL of RPMI cell
culture medium. Then, 15 mL of Histopaque was carefully layered Statistical Analysis
under the RPMI/blood mixture. After centrifugation (30 minutes at
Statgraphics Plus software (Statistical Graphics Corp, Rockville,
400 x g), the PMNC were removed and washed twice with phos-
MA) was used for all calculations. For the entire population, the
phate-buffered saline. Contaminating RBCs were hypotonically
gaussian distribution fitting of DPD activity was evaluated both
lysed. PMNC were suspended in 35 mmol/L sodium phosphate 2
graphically and statistically (X goodness of fit with a width interval
buffer containing 10% glycerol (25 x 106 cells/mL). The cell suspen-
of 0.030 nmol/min/mg protein, P = .10; Kolmogorov-Smirnov test,
sion was then centrifuged (5 minutes at 250 x g) and cells were
0 P = .29). Based on these results, which showed a satisfactory fit to
stored at -80 C (cell pellet plus supernatant) without impairment
the gaussian distribution, parametric statistical tests were used for
of enzymatic activity. 5
DPD activity. As we have previously shown, FU Cl and AUC did
not fit a gaussian distribution in the present study; thus, log Cl and
Determination of DPD Activity log AUC were used for statistical analysis. Two group comparisons
The day of the assay, the cell suspension was freeze-thawed three were performed using Student's t test. Comparison of DPD activity
times and centrifuged for 30 minutes at 28,000 x g (4°C). The between the 169 patients for whom DPD was investigated before or
supernatant was kept on ice until assayed (within 15 minutes) for at least 3 weeks after FU chemotherapy (mean, 0.222 nmol/min/mg
2250 ETIENNE ET AL

protein; range, 0.065 to 0.559) and the 16 patients who received FU 0.6
chemotherapy shortly before DPD investigation (mean, 0.227 nmoll
min/mg protein; range, 0.089 to 0.371) did not show any significant 0.5
difference (P = .82). In addition, intrapatient analysis confirmed that
a 5-day continuous infusion of FU did not influence DPD activity 0.4 a -II
measured 13 to 14 days after the end of the infusion: from cycle 1
to cycle 2, DPD activity decreased in nine patients and increased in 0.3 a . °1]* "
10 (Student's paired analysis on 19 patients, P = .62); and from cycle
00009
0
""a
2 to cycle 3, DPD activity decreased in 12 patients and increased in S a an ng
0.2
12 (Student's paired analysis on 24 patients, P = .32). The following a •
o
a
c-i a° a I•
analyses were thus performed on the entire set of patients: influence 0J
i
0.1
of age was analyzed by linear regression; influence of sex and hepatic
function was tested by multifactor analysis of variance, including a.
age as a covariate; and the correlation between FU C1 (log Cl) and
0 I-
DPD activity was tested by linear regression. 30 40 50 60 70 0o 90

Age
RESULTS
Fig 2. Plot of PMNC-DPD activity as a function of patient age (N
DPD Distribution = 185; linear correlation, P = .58).

The frequency histogram of PMNC-DPD for the entire

population (one measurement per patient) showed a uni-
modal distribution, which globally fits a gaussian distribu- influence DPD activity (Table 1). On average, DPD activ-
tion (Fig 1). Logarithmic transformation did not improve ity in women was 15% lower than in men (Table 1).
the normality of the distribution (data not shown). Mean This significant difference according to Student's t test
and median DPD activity values were 0.222 and 0.211 remained significant by multifactor analysis of variance
nmol/min/mg protein, respectively. A wide interpatient including hepatic function and age (Table 1). To evaluate
variability was observed (SD = 0.084; range, 0.065 to further the influence of hormonal status on DPD, we com-
0.559 nmol/min/mg protein). No total DPD deficiency pared DPD activity between premenopausal and post-
was found (ie, below the detection limit) in this large menopausal women and found no significant difference
group of unselected cancer patients. (mean, 0.203 and 0.191 nmol/min/mg protein, respec-
tively; P = .73).
Influence of Physiologic Parameters on DPD Activity
DPD Activity According to FU Pharmacokineticsand
Patient age did not significantly influence DPD activity Pharmacodynamics
(P = .58) (Fig 2). Abnormality of liver function did not
This analysis was performed on the 68 head and neck
cancer patients treated by neoadjuvant cisplatin/FU che-
motherapy (94 cycles analyzed). FU-related toxicity was
30 I-.' recorded for all cycles: hematologic toxicity (leukopenia
and/or thrombocytopenia grade 2 to 4) was observed in
nine cycles and digestive tract toxicity (mucositis grade
2 or 3) was observed in three other cycles. The risk of
5..
U
C

15
cr Table 1. Analysis of Influence of Sex and Hepatic
L
U_ Function on PMNC-DPD Activity
DPD Activity P

5 No, of (nmol/min/mg protein) Univariate Multivariate

Factor Patients Mean Range Analysis* Analysist

_1 Sex
o o.1 0.2 0.3 0.4 0.5 o.6 Male 152 0.228 0.065-0.559
Female 33 0.194 0.090-0.407
DPD activity (nmol/min/mg prot) Hepatic function
Normal 124 0.227 0.065-0.559
Fig 1. Frequency histogram with gaussian distribution fitting for Abnormal 58 0.212 0.080-0.459
PMNC-DPD activity in the entire population (N = 185). See Patients
and Methods for statistical evaluation of the gaussian distribution *Student's t test.
fitting. tMultifactor analysis of variance, including age as covariate.
DIHYDROPYRIMIDINE DEHYDROGENASE IN CANCER PATIENTS 2251

developing side effects was not linked to the pretreatment

DPD activity: means were 0.214 ± 0.082 and 0.200 +
0.075 nmol/min/mg protein for toxic and nontoxic cycles,
respectively (P = .56) (Fig 3). In contrast, FU systemic
exposure was linked to FU-related toxicity, with mean
AUCO-0 osh values being 33,673 and 25,268 ng/mL*h for
toxic and nontoxic cycles respectively (P = .0003) (Fig C
Oo
* 0 * oo
0

3). FU Cl showed a wide variability (mean, 2,251 mL/

min/m 2 ; range, 1,031 to 5,003). The correlation between L on
o o o

pretreatment DPD activity and FU Cl was poor (linear

regression [n = 90]: r = .31, P = .002) (Fig 4). Finally, | 0 U
we analyzed the link between the need for a pharmacoki- 0 0.1 0.2 0.3 0.4 0.

netically based FU dose reduction at midcycle and pre-

OPD activity (naol/min/ag prot)
treatment DPD activity (Table 2). Despite a tendency to
observe lower DPD activity in patients who required an Fig 4. Linear correlation between log 10 FU CI (y) and pretreat-
FU dose reduction, DPD values were not significantly ment PMNC-DPD activity (x): 90 cycles analyzed, y = 3.23 + 0.49 x,
r = .311, P= .0028.
different between cycles with dose reduction and those
without dose modification (P = .078).

DISCUSSION recent population study reported by Lu et al 22 attempted

to provide such data. The investigators demonstrated a
The recent literature has reported several cases of sus-
normal distribution of human PMNC-DPD on a set of
pectedis or proven12-14,16 DPD deficiencies associated with
124 subjects. These data are interesting, but unfortunately
severe FU-related toxicities, including death. The pending
the investigators examined healthy volunteers (staff and
and central question is to determine the incidence of DPD
students from the university), most of whom were
deficiency in a large population of cancer patients. A
younger than 50 years of age. Thus, their study suffered
first from the choice of the population itself (noncancer
A B subjects) and, second, from the age range, which was not
representative of cancer patients. These drawbacks limit
the possibility for extrapolation of their data to FU-treated
0.5
cancer patients. In the present prospective study, which
was performed on a large group of unselected cancer
patients, a normal distribution of PMNC-DPD activity
0.4 was also demonstrated. Interestingly, the mean value
0
e-
co
C-

at-
U
0
0.3 I 4
I
o

oo
found (0.222 nmol/min/mg protein) is close to that re-
ported by Lu et a122 (0.189 nmol/min/mg protein). It is
clear that no absolute DPD deficiency was identified by
either Lu et al or by ourselves. This means that total DPD
C0 0
0 deficiency is a rare event that is probably transmissible
-4 0.2 0 by an autosomal recessive pattern. 12 '13 Nevertheless, it
'I
0
must be underscored that low but still detectable DPD
I
0uo

0.1 I o
activity in PMNC is enough to cause severe FU-related
toxicities.16'
22

In a recent study based on more than 1,000 FU chemo-

0
I NT T NT T
Table 2. Analysis of Pretreatment PMNC-DPD Activity (nmol/min/mg
protein) According to FU Dose Modification Requirement
PMNC-DPD No Dose Modification Dose Reduction Dose Augmentation
Fig 3. (A) Plot of individual pretreatment PMNC-DPD activity in Activity (69 cycles) (22 cycles) (3 cycles)
FU-related toxic (T) and nontoxic (NT) cycles. (B) Plot of individual Mean 0.210 0.177 0.195
AUCo-10s h in FU-related T and NT cycles. Toxicity (hematologic and/ Median 0.200 0.140 0.221
or mucositis) was considered when ; grade 2 according to World
Range 0.080-0.414 0.089-0.362 0.101-0.264
Health Organization criteria.
2252 ETIENNE ET AL

therapy cycles,5 we demonstrated that FU Cl was signifi- body Cl of the drug is available. Such a strategy has been
cantly (12%) lower in women as compared with men adopted to approach optimal carboplatin dosage421 25 based
(mean, 2,733 and 3,100 mL/min/m 2 , respectively). From on the fact that carboplatin Cl is strongly correlated to
a preliminary study" based on 66 cancer patients (56 men glomerular filtration rate. It would be possible to tailor
and 10 women), we found that mean PMNC-DPD activity optimal FU dose based on PMNC-DPD activity providing
was lower in women (0.172 nmol/min/mg protein) than a strong correlation between FU Cl and PMNC-DPD ac-
in men (0.192 nmol/min/mg protein); due to the limited tivity. We previously reported a correlation between
size of the study population, this difference did not reach PMNC-DPD activity and FU Cl (r = .72).17 We reexam-
statistical significance. From the present investigation, ined this relationship in the present study on a larger set
which covered 185 patients, the average PMNC-DPD ac- of patients; although the correlation remained statistically
tivity value was 0.228 nmol/min/mg protein in men (n = significant, the link was found to be markedly weakened,
152) and 0.194 nmol/min/mg protein in women (n = 33); with the correlation coefficient decreasing to 0.31. Thus,
this difference was significant (P = .03). Interestingly, even though related to FU clearance, PMNC-DPD is not
this 15% difference in DPD activity is of the same order a sufficiently reliable predictor of FU Cl, and a DPD-
as the difference observed in FU C1. 5 Although FU is based FU dose individualization is not justified in our
mainly cleared in the liver,1 numerous arguments support opinion. Other investigators' opinions would be interest-
the fact that DPD activity measured in PMNC reflects ing on this point.
what occurs in the liver: severe FU-related toxicity re- Cisplatin induces anemia primarily and, to a lesser ex-
ported in patients with PMNC-DPD deficiency,12-14,16 the tent, leukopenia and thrombocytopenia. However, even
correlation reported between FU Cl and PMNC-DPD ac- though the presence of cisplatin adds to the hematologic
tivity,' 71' 8 and the link observed between hepatic and toxicity encountered, we have considered leukopenia and/
PMNC-DPD activity.22 We could hypothesize that the or thrombocytopenia mainly as related to FU. In
reduced capacity for FU Cl in women is based on a sex- agreement with our own experience and that of others,
related decrease in DPD activity. In the population study toxic cycles (ie, mucositis and/or leukopenia and/or
reported by Lu et al, 22 DPD activity was not influenced thrombocytopenia) were significantly related to increased
by sex. The discrepancy in the effect of sex on DPD FU systemic exposure (Table 2). The incidence of grade
activity between these two studies could be explained by 2 to 4 toxicity observed herein (13%) was similar to
the difference in age ranges covered, ie, hormonal factors that previously reported by us in FU-treated patients who
(premenopausal women in the majority in the study re- received identical schedules with individual pharmacoki-
ported by Lu et al v postmenopausal women in the major- netically based dose adaptation. 9 A corollary to the weak
ity in the present study). However, this hypothesis cannot correlation between FU Cl and PMNC-DPD activity was
be confirmed from our limited set of women, since no the fact that the necessity of individual dose reduction
difference in DPD activity was demonstrated between was not predictable on the basis of pretreatment DPD
premenopausal (n = 7) and postmenopausal (n = 26) determination (Table 2). Clearly, pretreatment DPD activ-
women. In agreement with Lu et al,22 we did not find any ity cannot be used to improve FU dose adaptation strat-
influence of age on DPD activity. This observation is in egy. However, based on previously reported cases,12-16
line with our previous results' which showed that age severe or total DPD deficiencies could be useful indica-
does not modify the capacity to clear FU. In a previous tors to avoid FU therapy in patients at risk. The difficulty
study,7 we also analyzed the influence of liver function is to set a lower limit of DPD activity to define such
and did not demonstrate a significant relationship between patients. Lu et a1 2 reported three new patients with FU
abnormality of the liver and FU Cl. The present data grade 5 toxicity who exhibited severe DPD deficiencies
confirm these previous observations, since patients with (range, 0.011 to 0.013 nmol/min/mg protein). We recently
liver abnormalities exhibited DPD activity comparable to examined two new cases of severe but reversible FU-
that of patients without hepatic dysfunction (Table 1). related toxicity (hematologic plus mucositis) in patients
Of importance is the clinical implication of individual who exhibited PMNC-DPD activity at 0.060 nmol/min/
knowledge of DPD activity, before starting FU treatment, mg protein' 6 and 0.090 nmol/min/mg protein (data not
in a cancer patient with planned FU treatment. For short published), respectively. Since the present study was per-
IV administration 23 and 5-day continuous infusion of formed on patients for whom individual FU dose adapta-
FU,s'9 an optimal full-cycle AUC has been identified. tion was systematically performed to control toxicity,9 the
Based on the following relationship: Dose = Cl x AUC, evaluation of a DPD threshold to identify patients at risk
the individual optimal dose can thus be determined before cannot be drawn from our data. Thus, based on the re-
starting treatment if a reliable means to evaluate total- ported cases from the literature,16 22 a risk threshold at
DIHYDROPYRIMIDINE DEHYDROGENASE IN CANCER PATIENTS 2253
0.100 nmol/min/mg protein can be proposed, below (ie, 2.7%) exhibited a DPD activity below this threshold.
which patients must be treated with markedly reduced We hope future prospective studies on PMNC-DPD activ-
FU doses or even an alternative chemotherapy regimen. ity in FU-treated patients will bring additional knowledge
The application of this threshold to the present population of the DPD activity range predictable for patients at risk
shows that five of 185 investigated unselected patients for impaired FU Cl and subsequent severe toxicity.

REFERENCES
1. Grem JL: Fluorinated pyrimidines, in Chabner BA, Collins JM toxicity secondary to dihydropyrimidine dehydrogenase deficiency:
(eds): Cancer Chemotherapy: Principles and Practice. Philadelphia, A potentially more common pharmacogenetic syndrome. Cancer
PA, Lippincott, 1990, pp 180-224 68:499-501, 1991
2. Grem JL, Shoemaker DD, Petrelli NJ, et al: Severe and fatal 15. Fleming RA, Milano GA, Gaspard MH, et al: Dihydropy-
toxic effects observed in treatment with high- and low-dose leuco- rimidine dehydrogenase activity in cancer patients. Eur J Cancer
vorin plus 5-fluorouracil for colorectal carcinoma. Cancer Treat Rep 29:740-744, 1993
71:1122, 1987 (letter) 16. Houyau P, Gay C, Chatelut E, et al: Severe fluorouracil toxic-
3. Merlano M, Grimaldi A, Brunetti I, et al: Simultaneous cis- ity in a patient with dihydropyrimidine dehydrogenase deficiency. J
platin and 5-fluorouracil as second-line treatment of head and neck Natl Cancer Inst 85:1602-1603, 1993
cancer. Cancer Treat Rep 71:485-488, 1987 17. Fleming RA, Milano G, Thyss A, et al: Correlation between
4. Lokich JJ, Ahlgren JD, Gullo JJ, et al: A prospective random- dihydropyrimidine dehydrogenase activity in peripheral mononu-
ized comparison of continuous infusion fluorouracil with a conven- clear cells and systemic clearance of fluorouracil in cancer patients.
tional bolus schedule in metastatic colorectal carcinoma: A Mid- Cancer Res 52:2899-2902, 1992
atlantic Oncology Program Study. J Clin Oncol 7:425-432, 1989 18. Harris BE, Song R, Soong S, et al: Relationship between
5. Milano G, Etienne MC, Cassuto-Viguier E, et al: Influence of dihydropyrimidine dehydrogenase activity and plasma 5-fluorouracil
sex and age on fluorouracil clearance. J Clin Oncol 10:1171-1175, levels with evidence for circadian variation of enzyme activity and
1992 plasma drug levels in cancer patients receiving 5-fluorouracil by
6. Goldberg JA, Kerr DJ, Willmott N, et al: Pharmacokinetics protracted continuous infusion. Cancer Res 50:197-201, 1990
and pharmacodynamics of locoregional 5-fluorouracil. Br J Cancer 19. Sommadossi JP, Gewirtz DA, Diasio RB, et al: Rapid catabo-
57:186-189, 1988 lism of 5-fluorouracil in freshly isolated rat hepatocytes as analyzed
7. Fleming RA, Milano G, Etienne MC, et al: No effect of dose, by high-performance liquid chromatography. J Biol Chem 257:8171-
hepatic function or nutritional status on FU clearance following 8176, 1982
continuous (5-days), FU infusion. Br J Cancer 66:668-672, 1992 20. Naguib FN, el Kouni MH, Cha S: Enzymes of uracil catabo-
lism in normal and neoplasic human tissues. Cancer Res 45:5405-
8. Thyss A, Milano G, Ren6ee N, et al: Clinical pharmacokinetic
study of 5-FU in continuous 5-day infusions for head and neck 5412, 1985
21. Christophidis N, Mihaly G, Vadja M, et al: Comparison of
cancer. Cancer Chemother Pharmacol 16:64-66, 1986
liquid- and gas-liquid chromatographic assays of 5-fluorouracil in
9. Santini J, Milano G, Thyss A, et al: FU therapeutic monitoring
plasma. Clin Chem 25:83-87, 1979
with dose adjustment leads to an improved therapeutic index in head
22. Lu Z, Zhang R, Diasio RB: Dihydropyrimidine dehydroge-
and neck cancer. Br J Cancer 59:287-290, 1989
nase activity in human peripheral blood mononuclear cells and liver:
10. Trump DL, Egorin MJ, Forrest A, et al: Pharmacokinetic
Population characteristics, newly identified deficient patients, and
and pharmacodynamic analysis of 5-fluorouracil during 72 hours clinical implication in 5-fluorouracil chemotherapy. Cancer Res
infusion with and without dipyridamole. J Clin Oncol 9:2027-2035,
53:5433-5438, 1993
1991 23. Van Groeningen CJ, Pinedo HM, Heddes J, et al: Pharmacoki-
11. Diasio RB, Harris BE: Clinical pharmacology of 5-fluoroura- netics of 5-fluorouracil assessed with a sensitive mass spectrometric
cil. Clin Pharmacokinet 16:215-237, 1989 method in patients on a dose escalation schedule. Cancer Res
12. Tuchman M, Stoeckeler JS, Kiang DT, et al: Familial pyrimi- 48:6956-6961, 1988
demia and pyrimidinuria associated with severe fluorouracil toxicity. 24. Calvert AH, Newell DR, Gumbrell LA, et al: Carboplatin
N Engl J Med 313:245-249, 1985 dosage: Prospective evaluation of a simple formula based on renal
13. Diasio RB, Beavers TL, Carpenter JT: Familial deficiency of function. J Clin Oncol 7:1748-1756, 1989
dihydropyrimidine dehydrogenase. Biochemical basis for familial 25. Egorin MJ, Van Echo DA, Tipping SJ, et al: Pharmacokinetic
pyrimidinemia and severe 5-fluorouracil-induced toxicity. J Clin In- and dosage reduction of cis-diamine (1,1-cyclobutanedicarbosylato)
vest 81:47-51, 1988 platinum in patients with impaired renal function. Cancer Res
14. Harris BE, Carpenter JT, Diasio RB: Severe 5-fluorouracil 44:5432-5438, 1984