Carboplatin Dosage: Prospective Evaluation of a Simple Formula

Based on Renal Function

By A.H. Calvert, D.R. Newell, L.A. Gumbrell, S. O'Reilly, M. Burnell, F.E. Boxall, Z.H. Siddik, I.R. Judson,
M.E. Gore, and E. Wiltshaw

A dosage formula has been derived from a retrospec-
tive analysis of carboplatin pharmacokinetics in 18
patients with pretreatment glomerular filtration rates
(GFR) in the range of 33 to 136 mL/min. Carboplatin
plasma clearance was linearly related to GFR (r = 0.85,
P < .00001) and rearrangement of the equation de-
scribing the correlation gave the dosage formula: dose
(mg) = target area under the free carboplatin plasma
concentration versus time curve (AUC) x (1.2 x GFR +
20). In a prospective clinical and pharmacokinetic
study the formula was used to determine the dose
required to treat 31 patients (GFR range, 33 to 135
mL/min) with 40 courses of carboplatin. The target
AUC was escalated from 3 to 8 mg carboplatin/mL/
min. Over this AUC range the formula accurately
predicted the observed AUC (observed/predicted ratio
1.24 ± 0.11, r = 0.886) and using these additional
data, the formula was refined. Dose (mg) = target
AUC x (GFR + 25) is now the recommended formula.
AUC values of 4 to 6 and 6 to 8 mg/mL • min gave rise
to manageable hematological toxicity in previously
treated and untreated patients, respectively, and
hence target AUC values of 5 and 7 mg/mL • min are
recommended for single-agent carboplatin in these
patient groups. Pharmacokinetic modeling demon-
strated that the formula was reasonably accurate
regardless of whether a one- or two-compartment
model most accurately described carboplatin pharma-
cokinetics, assuming that body size did not influence
nonrenal clearance. The validity of this assumption
was demonstrated in 13 patients where no correlation
between surface area and nonrenal clearance was
found (r = .31, P = .30). Therefore, the formula pro-
vides a simple and consistent method of determining
carboplatin dose in adults. Since the measure of
carboplatin exposure in the formula is AUC, and not
toxicity, it will not be influenced by previous or
concurrent myelosuppressive therapy or supportive
measures. The formula is therefore applicable to
combination and high-dose studies as well as conven-
tional single-agent therapy, although the target AUC
for carboplatin will need to be redefined for combina-
tion chemotherapy.
J Clin Oncol 7:1748-1756. © 1989 by American Soci-
ety of ClinicalOncology.

CARBOPLATIN was introduced in 1981 as
an analogue of cisplatin, which possessed
reduced nonhematological toxicities when com-
12
pared with cisplatin in experimental models. '
Early clinical studies established that carbo-
platin, when administered at the normal phase II
dose of 400 mg/m2, was virtually devoid of
nephrotoxicity, ototoxicity, and peripheral neuro-
toxicity. 3 5 The dose-limiting toxicity was to the
bone marrow with thrombocytopenia being more
marked than leukopenia. A large number of
From the Sections of Drug Development and Medicine,
Institute of Cancer Research and Royal Marsden Hospital,
London andSutton, Surrey, England.
Submitted January30, 1989; accepted June 7, 1989.
Supported by grantsfrom the CancerResearch Campaign
and MedicalResearch Council, England.
Address reprint requests to A.H. Calvert, MD, The
University of Newcastle upon Tyne, Division of Oncology,
Cancer Research Unit, Medical School, FramlingtonPlace,
Newcastle upon Tyne, NE2 4HH, England.
© 1989 by American Society of Clinical Oncology.
0732-183X/89/0711-0002$3.00/0
phase II and some phase III studies have been
performed with carboplatin, the results showing
that its antitumor activity and spectrum are
broadly similar to those of cisplatin.6
Although carboplatin is not significantly toxic
to the kidneys, it is clear that pretreatment renal
function markedly affects the severity of carbo-
platin-induced thrombocytopenia. In early stud-
ies it was noted that thrombocytopenia was more
prevalent in patients whose pretreatment glomer-
ular filtration rate (GFR) was reduced.87' Phar-
macokinetic studies provided a rationale for this
observation. The carboplatin molecule is stable
in plasma in vitro9-12 and remains so in vivo with
the majority, if not all, of the plasma platinum
being accounted for as the intact drug over the
initial 4- to 6-hour time period.9,13,14 Thereafter,
protein-bound platinum becomes the predomi-
nant species in the plasma. 9"' 3 18 However, as
platinum bound to plasma proteins is not cyto-
toxic, 19'2 0 pharmacokinetic studies have focused
on the fate of free intact carboplatin. The plasma
decay curves for intact carboplatin and free

1748 Journalof ClinicalOncology, Vol 7, No 11 (November), 1989: pp 1748-1756

Downloaded from ascopubs.org by DFG on December 17, 2016 from 130.133.008.114

Copyright © 2016 American Society of Clinical Oncology. All rights reserved.
CARBOPLATIN DOSAGE BASED ON RENAL FUNCTION
platinum following the administration of carbo-
platin are almost identical and, in the majority of
patients, can be fitted by a mono-exponential
curve,9,'13-15 although a rapid distribution phase
may also be detected.
Approximately 70% of an administered dose of
carboplatin is excreted in the urine and, when
chemical decomposition in the bladder is ac-
counted for, it seems probable that most is as the
intact drug.7-9 ' 13 -' 5"17' 2 1 22
, The renal clearance of
carboplatin is closely correlated with the GFR,
suggesting that the renal excretion of carboplatin
is accomplished exclusively by glomerular filtra-
tion. 9 Experimental studies in rats have con-
firmed that there is little tubular contribution to
the renal clearance of carboplatin and that less
than 2% of an administered dose is excreted in
the bile. 23 Carboplatin unaccounted for by renal
or biliary elimination is bound to tissues with the
fraction of the dose present in this form being
dependent upon the body size of the animal.
Thus, in mice and rats less than 10% of the
dose is unaccounted for by urinary excretion 23
while in patients the value is in the region of
30%.7-9,13-15,17,21,22
These relatively simple kinetics suggest that
the area under the concentration versus time
curve (AUC) for carboplatin (and consequently
the toxicity and therapeutic efficacy of the drug)
will be dictated primarily by the pretreatment
GFR and the dose administered with tissue
binding being the only other major determinant
of disposition.
In view of this, and in the light of clinical
experience, several authors have recommended
adjustment of the carboplatin dose depending on
renal function. Egorin et a18 showed that the
thrombocyte nadir (expressed as the percentage
of the pretreatment count) was correlated with
the free platinum AUC, following the administra-
tion of intravenous (IV) carboplatin. These au-
thors derived a formula from which the dose of
carboplatin necessary to induce a chosen throm-
bocyte nadir could be calculated for an individual
patient from the pretreatment GFR, the surface
area, and the patient's prior therapy status.
Although its predictive accuracy has been ques-
tioned,24 this formula was validated in a prospec-
tive study. 25 Taguchi et a126 have also shown that
impaired renal function is associated with a
Downloaded from ascopubs.org by DFG on December 17, 2016 from 130.133.008.114
Copyright © 2016 American Society of Clinical Oncology. All rights reserved.
1749
greater degree of thrombocytopenia and have
suggested an empirical dosage scheme.
A retrospective study of our own phase I and
pharmacokinetic data showed that the free carbo-
platin AUC was linearly related to dose only
when allowance was made for variations in
GFR. 27 This observation suggests that phase II
doses (usually 400 mg/m 2), which had been
established mainly on patients with filtration
rates in the region of 60 to 100 mL/min, would
be inadequate (by as much as twofold) when
administered to patients with higher GFRs.
Therefore, the aim of the present study was to
develop a formula by which the dose of carbo-
platin necessary to achieve a given AUC could be
calculated. AUC, rather than a given thrombo-
cyte nadir, was chosen as the end point for the
following reasons: (1) the development of drug
combinations containing carboplatin demands a
method by which the patient's exposure to carbo-
platin may be kept constant, although the blood
count nadirs will be affected by the other compo-
nents of the combination; (2) when high-dose
carboplatin is administered with hematological
support the blood count nadirs are no longer
meaningful. A method to predict AUC in these
patients will allow consistent drug exposures to
be achieved and the existence of a dose-response
relationship to be investigated; and (3) the use of
AUC allows a precise end point to be used as the
basis for the model development. Thrombocyte
nadirs are time-dependent, are affected by previ-
ous treatment, and can only be determined accu-
rately with very close patient monitoring. Cer-
tain aspects of this study have been reported
28 29
before in abstract form. ,
METHODS
The current study was conducted in three stages. First, a
dosage formula was derived retrospectively from our previous
data collected during phase I and II pharmacokinetic studies
(part I). Second, this formula was used to determine the doses
used in a prospective clinical/pharmacokinetic study in
which the AUC was escalated to validate the dose equation
derived in part I and to determine the AUC/toxicity relation-
ship (part II). Third, the new data derived from the second
stage were used to refine the dosage formula (part III) and
model-dependent pharmacokinetic analyses were performed
to investigate the effect of one- and two-compartment pharma-
cokinetic models on the predictive accuracy of the formula
derived. Ethical permission was obtained for all clinical and
pharmacokinetic studies from the Ethical Committee of the
Royal Marsden Hospital.
1750
Part I
The initial carboplatin dosage formula was developed
using pharmacokinetic data published previously., 27 The
characteristics of these patients are given in Table 1.
PartII
The characteristics of the patients who were studied in part
II of the investigation are also summarized in Table 1. GFR
was estimated by the 5" CrEDTA method. 30 Full blood counts
were obtained before treatment and 3 weeks after treatment
in patients who received carboplatin at doses designed to
achieve AUC values of 3 to 5 mg carboplatin/ml • min. In
patients treated with carboplatin at doses designed to achieve
6 to 8 mg carboplatin/mL/min an additional blood count was
performed, when possible, 2 weeks after treatment. Carbo-
platin (Bristol-Myers, Wallingford, CT) was administered
over 1 hour in a 500 mL 5% dextrose solution. The dose
administered was derived from the formula developed in part
I: Dose of carboplatin (mg) = Target AUC x
(1.2 x GFR + 20). In this formula, and throughout this
report, the units for the carboplatin AUC are mg carboplatin/
mL • min and not mg elemental platinum. The target AUC
was escalated from 3 to 8 mg/mL • min and the number of
patients treated and courses studied at each AUC level are
given in Table 2. The initial target AUC (3 mg/mL - min)
and subsequent increments (1 mg/mL • min) were based
7 9 27
upon the results obtained in our earlier clinical studies. ,'
Pharmacokinetic studies were performed in all patients to
determine the actual AUC achieved by collecting blood
samples from an indwelling IV cannula placed in the arm
opposite to the one receiving the carboplatin infusion. The
patency of the cannula was maintained by flushing with 10
IU/mL heparin. Blood samples were withdrawn prior to the
start of the carboplatin infusion, at midinfusion, at the end of
the infusion, and at the following time points thereafter: 0.25,
0.5, 1, 1.5, 2, 4, 6, 9, 12, 18, and 24 hours. Plasma was
prepared within 15 minutes by centrifugation at 40C and then
plasma ultrafiltrates were prepared immediately as described
previously using Amicon Centrifree (Stonehouse, Gloucester-
shire, UK) micropartition units.9"1 8 The ultrafiltrates of
plasma were then stored at - 20 0C until analysis of plasma-
Table 1. Characteristics of Patients Studied
Part I Part If
Number 18 31
M/F 11/7 4/27
Mean age (range) 55 (24-78) 57 (17-73)
GFR (range) 33-136 mL/min 33 to 135 mL/min
Carboplatin dose 150/500 mg/m2
Diagnosis Ovary 7
Lung 3
Mesothelioma 2
Testicular 3
Head and neck 2 Ovary 24
Pancreas 1 Sarcoma 7
Prior chemother-
apy 17
No prior chemo-
therapy 14
Downloaded from ascopubs.org by DFG on December 17, 2016 from 130.133.008.114
Copyright © 2016 American Society of Clinical Oncology. All rights reserved.
CALVERT ET AL
free platinum levels was determined using flameless atomic
absorption spectrophotometry.9 '8 The plasma-free platinum
AUC, expressed as mg/mL • min of carboplatin, was
calculated by the trapezoidal rule. The time 0 plasma
carboplatin concentration was taken as 0 and no correction
for the period from the final time point to infinity was
employed.
PartIII
Using the data set generated in part II, a new dosage
formula was derived using the simple one-compartment
model analysis previously described. 27 Although such a model
fit the majority of our data, particularly over the 0- to 8-hour
time period in which the majority of the AUC is defined, we
also investigated a two-compartment model to see whether
significant errors would be introduced if such a model were
operative. Two such models were considered. In the first, the
nonrenal elimination of carboplatin was assumed to be from
the peripheral compartment, and in the second to be from the
central compartment. Analytical solutions were obtained
from published literature" and by standard methods. 32 It was
possible to refer to the analytical solutions to investigate the
importance of body size"; numerical examples were used to
assess the possible inaccuracy that might be introduced by
the assumption of an inverse linear relationship between
AUC and GFR.
RESULTS
Part I: Derivationof the Initial Carboplatin
Dose Formula
Figure 1 shows the correlation of free carbo-
platin plasma clearance with GFR in 18 patients
treated with carboplatin as part of the initial
clinical evaluation of the compound. The equa-
tion describing the correlation: Clearance =
Dose/AUC = 1.21 (± 0.19) x GFR + 23
(+ 16) (figures in parentheses are SEM) was
rearranged and the constants rounded to give the
initial carboplatin dose formula: Dose (mg) =
AUC x (1.2 x GFR + 20).
PartII: Prospective Evaluation of the Dosage
Formula
Pharmacokinetics. The dose formula de-
rived in part I was then evaluated in 31 patients
who received 40 courses of carboplatin. Figure 2
shows the correlation of the predicted AUC with
that actually measured and Table 2 shows the
actual doses given to the patients and their renal
function. During the course of the study the
method of reporting 51CrEDTA clearances used
in our hospital was changed. Initially, the actual
measured value was reported, but later the re-
ported values were normalized for a body surface
area of 1.73 m2. This meant that some of our
doses were inadvertently calculated using the
CARBOPLATIN DOSAGE BASED ON RENAL FUNCTION 1751

28e
lllr117111111111111 lllrlll11111
12

ii11
E E
Q / A 18
E
V 16g -4
0 / E
X 9
IS
0El
U 2
U
O 8
S12S U
U
7
C
iee -
.4
6
V
6
E 7 4
5
U 7
7S I I .. I
U o
U L
13
J3
C 4
-.4
4.)' 10
*aeBe
5e iN
3
0.
oJ3
.0 L
L
LU 2 a
U 2
25 -

0 .0
±

e
0 25 56 75 1GO 125 158
0 1 2 3 4 5 6 7 8 9 18 11 12
GFR (ml/min)
Predicted Curboplatin AUC (mg/mixmin)

Fig 1. Correlation of free carboplatin plasma clearance

with pretreatment GFR in patients studied to derive the
initial carboplatin dose formula. The solid line is the linear
regression line and the broken lines the 95% confidence
intervals. Mean + SE for the slope = 1.21 ± 0.19 and for the
y intercept = 23 ± 16. Correlation coefficient r = 0.851, P =
< .00001.
Fig 2. Correlation of observed carboplatin AUC with
predicted carboplatin AUC following dosage according to the
formula: Dose (mg) = Target AUC x (1.2 x GFR + 20). The
solid line is the linear regression line and the broken lines are
the 95% confidence intervals. Mean ± SE for the slope =
1.24 ± 0.11 and for the intercept -0.49 ± 0.58. Correlation
coefficient r = 0.886, P < .00001.

normalized rather than the absolute clearance.

We have compensated for this error by making a
proportionate adjustment in the predicted AUC
plotted in Fig 2.
From Fig 2 it is clear that there was a good
correlation between the predicted and observed
AUC (r = 0.886, P = < 0.00001), despite the
wide variation in renal function and doses (ex-
pressed as mg/m2) administered (Table 2). For
example, at a predicted AUC of 4 mg/mL . min
the variation in the actual dose administered was
3.3-fold whereas the variation in the ratio of the
predicted to the observed AUC was only 1.5-fold.
However, at higher target AUC values, notably 8
mg/mL • min, the narrower range of GFR
values resulted in little difference in the spread of
dose and AUC values. Despite the good correla-
tion between the observed and predicted carbo-
platin AUC, the formula consistently under-
predicted by about 20% (the gradient of the
regression line being 1.24 ± 0.11), demonstrat-
ing the need to revise the formula for subsequent
use. This revision constituted part III of the
study.
Thrombocytopenia. Table 3 shows the plate-
let counts observed in patients treated during
part II of the present study; data are given for the
first course of therapy only. There was a reduc-
tion in the platelet count in all patients with the
more severe thrombocytopenia occurring at the
higher AUC values, and counts at 2 weeks after
treatment were, in general, lower than at 3
weeks. In all patients where data are available
(n = 21), platelet counts recovered to at least
60% of the pretreatment value by 4 weeks after
treatment with the absolute counts being greater
than 100,000 x 106/L in every case. In four of 21
cases, platelet nadirs of less than 100,000 x
106/L were recorded (World Health Organisa-
tion [WHO] grade of thrombocytopenia > 0).
Two chemotherapy-naYve patients who achieved
AUC values of 11.7 and 9.9 mg/mL - min
developed platelet nadirs of 15% and 18% (WHO
grades 1 and 2) of the pretreatment value. A
third previously untreated patient achieved an
AUC of 7.7 mg/mL • min and experienced
grade 2 thrombocytopenia (7% pretreatment);

Downloaded from ascopubs.org by DFG on December 17, 2016 from 130.133.008.114

Copyright © 2016 American Society of Clinical Oncology. All rights reserved.
1752 CALVERT ET AL

Table 2. Carboplatin AUC Escalation

AUC Ratio*
Target AUC GFR* Dose* Observed/
2
(mg/mL - min) No. Patients No. Courses (mL/min) (mg/m ) Predicted
3 5 7 33-87 94-197 .99-1.10
4 5 6 37-119 125-412 .93-1.39
5 8 8 40-118 286-531 .96-1.29
6 11 11 49-135 281-642 .94-1.62
7 5 5 63-113 417-633 .90-1.48
8 3 3 60-77 423-581 .99-1.46
*Range of values.

however, this patient had a deep vein thrombosis,
suggesting the possibility of platelet consump-
tion. A final patient (patient A), who achieved an
AUC of 7.7 mg/mL - min, developed grade 3
thrombocytopenia (6% pretreatment), probably
as a result of her prior treatment with six courses
of doxorubicin, cyclophosphamide, and cisplatin
combination therapy.
Leukopenia. Only one previously treated pa-
tient (patient A) experienced a greater than 50%
reduction in the white cell count following the
first course of therapy. In chemotherapy-naive
patients, a 50% to 80% reduction in the leukocyte
count was observed in five of 13 patients 3 weeks
after treatment. This represented WHO grade 1
toxicity in four cases and grade 2 toxicity in the
other patient.
Other toxicities. The only other toxicity seen
in this study was mild nausea and vomiting,
which were well controlled by standard antiemet-
ics in all cases.
PartIII: Revision of the Dosage Formula and
Investigation of the Model Dependency of the
Formula
Figure 3 is a plot of carboplatin plasma clear-
ance (dose/AUC) versus GFR for the patients
investigated in part II of the study. The graph is
similar to that derived retrospectively from pa-
tients evaluated in initial clinical studies with
carboplatin (Fig 1) except that the gradient is
closer to unity. Therefore, the revised dosage
formula derived from the regression analysis of
these data is: Dose (mg) = AUC x (GFR + 25).
This dose formula is consistent with a single-
compartment pharmacokinetic model for phar-
macokinetics. Although there is now a large body
of data demonstrating that following a rapid
distribution phase, plasma free carboplatin/
platinum levels can be described by a one-
compartment model for up to 6 to 8 hours after
administration and possibly longer, 9"1318 ,21,22 we
also investigated whether the operation of a
two-compartment model would introduce signifi-
cant errors in the AUC prediction.
In the initial solution of the one-compartment
model, the renal carboplatin clearance was as-
sumed to be proportional to the GFR.27 The
nonrenal clearance for these patients was mea-
sured as the difference between the renal clear-
ance and the total plasma clearance. No correla-
tion was found between this nonrenal clearance
and body surface area (Fig 4) so the nonrenal
clearance was assumed to be constant. The solu-
tion of the model yielded the formula: Dose
(mg) = AUC x (A x GFR + B) where A is the

Table 3. Thrombocytopenia Following the First Course of Carboplatin

Thrombocytopenia*

Observed AUC Prior Observed

% AUCPror
Pretreatment Thrombocyte Count (x 10'/L)
(mg/mL - min) No. Patients Chemotherapy 2 Wk 3 Wk 2 Wk 3 Wk
<4 2 2/2 - 51%,76% - 121,191
4-6 5 5/5 27% 45% (36-83%) 103 164 (110-189)
6-8 9 1/9 18% (6-58%) 61% (36-73%) 113 (21-139) 237 (153-484)
>8 5 0/5 15, 18, 26% 81% (63-110%) 51, 85, 100 302 (244-637)
*Thrombocyte counts are expressed as both % pretreatment and as the absolute counts observed. For n > 5 observations the median and
range of values are given, and where n < 5 individual counts are given. Although 3-week blood counts were not available for all patients whose
target AUC was 3, 4, or 5 mg/mL - min, counts are given for all patients whose first course target AUC was 6 mg/mL
t min or greater. The only
exception was a patient who died unexpectedly of rapidly progressive disease before reaching the nadir.

Downloaded from ascopubs.org by DFG on December 17, 2016 from 130.133.008.114

Copyright © 2016 American Society of Clinical Oncology. All rights reserved.
CARBOPLATIN DOSAGE BASED ON RENAL FUNCTION 1753

P'1' I"T'1 ' ' ' ' i ' ' ' ' I ' ' ' ' I ' ' ' 'r
175 ±16

'.4
E 5

80
U
6 125
E

u
p

C
II
U L
I-L
r
E 100
II H
a 6B
o
"75 L.0
C
O.
9
U
34 C
6 56e
L
LI 40 SD
0
.1 0 jJo S
25
L Z

0 20

a 25 B5 75 18 125 150 1.4 1.5 1.6 1.7 1.8 1.9 2

QFR (ml/min) Surface Area (Sqr.m)

Fig 3. Correlation of free carboplatin plasma clearance Fig 4. Relationship between nonrenal clearance and
with pretreatment GFR in patients studied during the body surface area in the 13 patients studied retrospectively.
prospective validation of the dose formula derived in Part I. The -- is that given by linear regression analysis and the
The - - is the linear regression line and the - are the 95% -- the 95% confidence intervals (r = 0.312, P = .30).
confidence intervals. Mean ± SE for the slope = 0.93 ± 0.08
and for the y intercept 26 + 6. Correlation coefficient r =
0.875, P= <.00001. achieved when an individual patient is given
carboplatin using a simple dosage formula in
ratio of GFR to the renal clearance of carbo- which the only variable is renal function. The
platin and B is the nonrenal clearance. pharmacokinetically derived dosage formula may
This solution implies that the dose required to
produce a particular AUC in the patient is Central K Peripheral Model A
Compartment Compartment
linearly dependent on the GFR but is indepen- tacelular fluid)
dent of any measure of the patient's size, a result Renal Tissue
that is rendered less surprising by the knowledge Excretion Binding

that, in the normal population, GFR tends to

vary in proportion to body surface area.
The two further models studied in order to C..m Pemiphersl Model B
establish whether the existence of a two-compart- (excrsoellulsr fluid) Kv
ment system would significantly affect the dose
Renal Plasma
predictions are shown in Fig 5. Excretion Protein
Binding
It was possible to demonstrate analytically
that the AUC obtained was independent of
compartment size, providing that the nonrenal GFR 120 60 30 GFRVaried

clearance was constant.31 It was also possible to AUC so69 131 232 Weight 70kg rudican

140 Weight Varied dousing

demonstrate numerically that the AUC was Weight 35 70

approximately linearly related inversely to GFR AUC 131 131 131 GFR 60ml/mln

with the values estimated for A and B.

Fig 5. Two possible two-compartment models used to
DISCUSSION represent carboplatin disposition. The nonrenal component is
(a) placed in either the peripheral compartment, (to simulate
The results described in this report show that it tissue binding), or (b) in the central compartment (to simulate
is possible to predict the AUC that will be plasma protein binding).

Downloaded from ascopubs.org by DFG on December 17, 2016 from 130.133.008.114

Copyright © 2016 American Society of Clinical Oncology. All rights reserved.
1754
be used to minimize the variations in the AUC in
different patients by taking account of their renal
function. If one considers the use only of the
conventional mg/m 2 basis for adult patients,
considerably greater variations in the AUC (and
thus the exposure to the toxic and therapeutic
effects of the drug) may be expected. In our
experience, the actual GFR of patients with
overtly "normal" renal function may vary from
60 mL/min (or less) to 180 mL/min. If the AUC
required was 5 mg/mL - min the doses predicted
for two such patients would be 425 and 1025 mg,
respectively. If we assume that the same patients
both had a body surface area of 1.7 m 2 and were
given 400 mg/m2, then the actual dose given
would be 680 mg, which would be expected to
result in AUCs of 8 and 3.3, respectively, suggest-
ing substantial under-dosage for the high-
clearance patient. The idea of reducing the dose
of a drug in patients with poor renal function is
well accepted. The current data also suggest that
increasing the dose in patients with "normal" but
higher-than-average clearances will be necessary
to exploit the full therapeutic potential of carbo-
platin.
Although the central aim of the present study
was to validate and refine the carboplatin dose
formula, the toxicity data obtained do allow
recommendations to be made for the carboplatin
AUC values that can be safely achieved when the
drug is used as a single agent. In previously
treated patients, AUC values within the range of
4 to 6 mg/mL - min were well tolerated, whereas
in previously untreated patients, AUC values in
the range of 6 to 8 mg/mL - min were associated
with manageable thrombocytopenia. On the ba-
sis of these data, 5 and 7 mg/ml • min are
recommended as target AUC values for previ-
ously treated and untreated patients, respec-
tively. In combination therapy, the appropriate
AUC value depends upon the other drugs con-
cerned. For example, in the combination of
etoposide, bleomycin, and carboplatin (for testic-
ular teratoma) a value of 4.5 mg/mL - min has
been used with acceptable and predictable myelo-
suppression even though the actual doses given
33
were as high as 900 mg.
Retrospective analysis of data obtained from
patients treated with 800 to 1600 mg/m2
carboplatin1 8 indicates that the AUC predicted
by the revised dosage formula is in all cases
within 20% of the observed AUC with 28 mg/
Downloaded from ascopubs.org by DFG on December 17, 2016 from 130.133.008.114
Copyright © 2016 American Society of Clinical Oncology. All rights reserved.
CALVERT ET AL
mL - min being the highest AUC achieved. The
dose prediction formula should also be of value in
high-dose carboplatin studies.
Both the one- and two-compartment model
analyses described here predicted that a measure
of body size (weight or surface area) would be
unnecessary, providing that the nonrenal clear-
ance remained constant between patients. The
results obtained on adult patients reported here
seem to vindicate the prediction. Nevertheless,
since the nonrenal clearance is almost certainly
due to irreversible tissue binding, it is reasonable
to suppose that it will vary with body size. The
formula derived here predicts reasonably accu-
rately, probably because the nonrenal component
of elimination is small (approximately 30% in
patients with average renal function) compared
with the total and the variations between the
weights of the adults studied were only about
twofold. One could expect errors of about 10% in
the predicted dose for adults. Because of these
considerations and the difficulty in obtaining a
clear relationship between body size and AUC
from the retrospective data, we felt that the
assumption of invariant nonrenal clearance was
reasonable for the adult population. If children
are to be treated, particularly if they have im-
paired renal function, the assumption of constant
nonrenal clearance might introduce greater er-
rors. For this reason, we have started studies to
measure directly the nonrenal clearance in chil-
dren and to correlate this with some measure of
their body size.
The carboplatin dose formula derived by Egorin
et a18 took account of surface area but also
predicted the dose to be administered in terms of
mg/m2. The equation described by these authors
may thus be rearranged to provide a formula that
gives the absolute dose to be administered. This
rearranged formula has two components, a large
one dependent upon renal function and a small
one dependent upon surface area. These results
confirm that the body surface area of adult
patients is of relatively little significance in deter-
mining the dose of carboplatin that should be
given to adults, providing that an accurate esti-
mate of GFR is available.
In the present study, GFR has been measured
by the method of 5 1CrEDTA clearance. 30 The
advantage of this technique is that it allows
direct measurement of GFR to be made, in
contrast to creatinine clearance where other
CARBOPLATIN DOSAGE BASED ON RENAL FUNCTION
factors can influence serum creatinine levels and
hence the calculated GFR. In a recent study,
Daugaard et a134 have shown that prior to the
initiation of cisplatin therapy 51CrEDTA and
creatinine clearances correlate well with a ratio
close to one (0.91). However, once patients had
been treated with cisplatin there was no relation-
ship between the true GFR (as measured by
51
" CrEDTA) and creatinine clearance. Although
it would be possible to use pretreatment GFR as
measured by creatinine clearance to determine
the correct dose of carboplatin, any subsequent
adjustment in the face of altered renal function
would require the use of an isotopic method.
Similarly, in patients who have previously re-
ceived nephrotoxic chemotherapy, creatinine
clearance may not represent a valid measure of
GFR. These considerations make "5 CrEDTA
clearance, or similar techniques, the method of
choice for the determination of GFR for carbo-
platin dose determination.
Collins et al have shown that when various
cytotoxic drugs are given to different species at
an LD 10 or maximum-tolerated dose, the plasma
AUC tends to be more or less constant across the
species."s This is particularly true in the case of
platinum complexes.3 6 If it is assumed that nonre-
nal clearance is directly proportional to body
weight, then for both mice and rats the nonrenal
component of carboplatin clearance can be ig-
nored since it will be small (ie, for a 20 g mouse
nonrenal clearance will be 0.02/70 that of a
human, and for a 200 g rat 0.2/70 that of a
human). The carboplatin dose formula then
simplifies to: Dose (mg) = Target AUC x GFR.
For mice, the LD10 of carboplatin is 165 mg/
kg 36 and the GFR 27 mL/min/kg (Jodrell and
Newell, personal communication, 1988) and
therefore, the AUC at the LD10 is calculated as
6.1 mg/mL - min. For rats, the MTD is 60
mg/kg and the GFR 10 mL/min/kg2 3 and hence,
the calculated AUC is 6.0 mg/mL . min. These
values agree well with those observed in patients,
and the ability of the dosage model to extrapolate
so precisely across species so variable in size
REFERENCES
1. Harrap KR: Preclinical studies identifying carboplatin
as a viable cisplatin alternative. Cancer Treat Rev 12:21-33,
1985
2. Rose WC, Schurig JE: Preclinical antiturnour and
toxicologic profile of carboplatin. Cancer Treat Rev 12:1-19,
1985
Downloaded from ascopubs.org by DFG on December 17, 2016 from 130.133.008.114
Copyright © 2016 American Society of Clinical Oncology. All rights reserved.
1755
supports the idea that the pharmacokinetic as-
sumptions made here are reasonable.
The clear relationship between dose and toxic-
ity for carboplatin has been described in numer-
ous phase I investigations, and the relationship
between AUC and thrombocytopenia has been
described in other studies.8 26 The justification for
the use of pharmacokinetically guided dosage
regimens to increase the therapeutic efficacy of
carboplatin relies on the idea that there is also a
dose-response (or AUC-response) effect. Labora-
tory evidence, both in animal tumors and in
human tumor xenografts, suggests that this is
indeed the case,"' 2 and certain clinical studies
also suggest a dose-response effect. For example,
a randomized study in ovarian cancer in which a
low dose (150 mg/m2) was used showed a poor
result for carboplatin compared with cisplatin,3 7
while studies in which 400 mg/m2 has been used
show similar efficacy. 38 We plan to undertake
studies comparing two AUC values to further
investigate this point. It is hoped that, as has
recently been shown for teniposide, 39 an exposure-
efficacy relationship may become apparent.
To conclude, this study describes the deriva-
tion and prospective validation of a simple for-
mula for the calculation of carboplatin dosage in
adults. The formula compensates for variations
in pretreatment renal function that might other-
wise result in either underdosing (in patients with
above average renal function) or overdosing (in
patients with impaired renal function). As carbo-
platin plasma AUC, rather than toxicity, is the
measure of drug exposure defined by the for-
mula, it can be applied to both combination and
high-dose therapy. It is recommended that future
studies with carboplatin be conducted with the
dose determined by the formula described here to
ensure that the full therapeutic potential of this
drug is realized.
ACKNOWLEDGMENT
The authors gratefully acknowledge the medical and
nursing staff of the Royal Marsden Hospital for their skilled
care of the patients who were treated as part of this study.
3. Canetta R, Rozencweig M, Carter SK: Carboplatin:
The clinical spectrum to date. Cancer Treat Rev 12:125-136,
1985
4. Foster BJ, Clagett-Carr K, Leyland-Jones B, et al:
Results of NCI-sponsored phase I trials with carboplatin.
Cancer Treat Rev 12:43-49, 1985
1756
5. Van Glabbeke M, Renard J, Pinedo HM, et al: Ipropla-
tin and carboplatin induced toxicities: Overview of phase II
clinical trials conducted by the EORTC Early Clinical Trials
Cooperative Group (ECTG). Eur J Cancer Clin Oncol
24:255-262, 1988
6. Canetta R, Bragman K, Smaldone L, et al: Carboplatin:
Current status and future prospects. Cancer Treat Rev
15:17-32, 1988
7. Calvert AH, Harland SJ, Newell DR, et al: Early
clinical studies with cis-diammine-1,1-cyclobutane dicarbox-
ylate platinum II. Cancer Chemother Pharmacol 9:140-147,
1982
8. Egorin MJ, Van Echo DA, Tipping SJ, et al: Pharma-
cokinetics and dosage reduction of cis-diammine(l,1-cyclo-
butanedicarboxylato)platinum in patients with impaired re-
nal function. Cancer Res 44:5432-5438, 1984
9. Harland SJ, Newell DR, Siddik ZH, et al: Pharmaco-
kinetics of cis-diammine- 1,1 -cyclobutane dicarboxylate plat-
inum(II) in patients with normal and impaired renal func-
tion. Cancer Res 44:1693-1697, 1984
10. Momburg R, Bourdeaux M, Sarrazin M, et al: In vitro
plasma binding of some second generation antitumour plati-
num complexes. Eur J Drug Metab Pharmacokinet 10:77-83,
1985
11. Gaver RC, Deeb G: High performance liquid chromato-
graphic procedures for the analysis of carboplatin in human
plasma and urine. Cancer Chemother Pharmacol 16:201-
206,1986
12. van der Vijgh WJF, Klein I: Protein binding of five
platinum compounds: comparison of two ultrafiltration sys-
tems. Cancer Chemother Pharmacol 18:129-132, 1986
13. Elferink F, van der Vijgh WJF, Klein I, et al:
Pharmacokinetics of carboplatin after iv administration.
Cancer Treat Rep 71:1231-1237, 1987
14. Oguri S, Sakakibara T, Mase H, et al: Clinical
pharmacokinetics of carboplatin. J Clin Pharmacol 28:208-
215, 1988
15. Gaver RC, Colombo N, Grenn MD, et al: The
disposition of carboplatin in ovarian cancer patients. Cancer
Chemother Pharmacol 22:263-270, 1988
16. Curt GA, Grygiel JJ, Corden BJ, et al: A phase I and
pharmacokinetic study of diamminecyclobutane-dicarbox-
ylatoplatinum (NSC 241240). Cancer Res 43:4470-4473,
1983
17. Van Echo DA, Egorin MJ, Whitacre MY, et al: Phase
I clinical and pharmacological trial of carboplatin daily for 5
days. Cancer Treat Rep 68:1103-1114, 1984
18. Newell DR, Siddik ZH, Gumbrell LA, et al: Plasma
free platinum pharmacokinetics in patients treated with high
dose carboplatin. Eur J Cancer Clin Oncol 23:1399-1405,
1987
19. Gormley PE, Bull JM, LeRoy AF, et al: Kinetics of
cis-dichlorodiammineplatinum. Clin Pharmacol Ther 25:351-
357, 1979
20. Takahashi K, Seki T, Nishikawa K, et al: Antitumour
activity and toxicity of serum protein-bound platinum formed
from cisplatin. Jpn J Cancer Res (GANN) 76:68-74, 1985
21. Koeller JM, Trump DL, Tutsch KD, et al: Phase I
clinical trial and pharmacokinetics of carboplatin (NSC
241240) by single monthly 30-minute infusion. Cancer 57:222-
225, 1986
Downloaded from ascopubs.org by DFG on December 17, 2016 from 130.133.008.114
Copyright © 2016 American Society of Clinical Oncology. All rights reserved.
CALVERT ET AL
22. Reece PA, Bishop JF, Oliver IN, et al: Pharmacokinet-
ics of unchanged carboplatin (CBDCA) in patients with
small cell lung carcinoma. Cancer Chemother Pharmacol
19:326-330, 1987
23. Siddik ZH, Newell DR, Boxall FE, et al: The compar-
ative pharmacokinetics of carboplatin and cisplatin in mice
and rats. Biochem Pharmacol 36:1925-1932, 1987
24. Fish RG, Adams M, Shelley MD, et al: Validity of a
dosing scheme for carboplatin. Proc Am Soc Clin Oncol 6:21,
1987 (abstr 76)
25. Egorin MJ, Van Echo DA, Olman EA, et al: Prospec-
tive validation of a pharmacologically based dosing scheme
for the cis-diamminedichloroplatinum(II) analogue di-
amminecyclobutane-dicaboxylatoplatinum. Cancer Res 45:
6502-6506, 1985
26. Taguchi J, Saijo N, Miura K, et al: Prediction of
haematologic toxicity of carboplatin by creatinine clearance
rate. Jpn J Cancer Res 78:977-982, 1987
27. Calvert AH, Harland SJ, Newell DR, et al: Phase I
studies with carboplatin at The Royal Marsden Hospital.
Cancer Treat Rev 12:51-57, 1985
28. Calvert AH, Newell DR, Gumbrell LA, et al: Carbo-
platin: Prospective pharmacokinetically guided dose-escala-
tion in relation to renal function. Proc Am Soc Clin Oncol
6:45, 1987 (abstr 172)
29. Newell DR, Calvert AH, O'Reilly S, et al: Prospective
validation of a simple formula for determining carboplatin
dosage. Br J Cancer 56:233, 1987 (abstr)
30. Chantler C, Garnett ES, Parsons V, et al: Glomerular
filtration rate measurement in man by the single injection
method using 5 1Cr-EDTA. J Clin Sci 37:169-190, 1969
31. Welling PG: Pharmacokinetics, Processes and Mathe-
matics. American Chemical Society Monograph 185, 1986
32. Piaggio HTH: An Elementary Treatise on Differential
Equations and Their Applications. London, England, G Bell,
1928, pp 25-48
33. Newell DR, Eeles RA, Gumbrell LA, et al: Carbopla-
tin and etoposide pharmacokinetics in patients with testicular
teratoma. Cancer Chemother Pharmacol 23:367-372, 1989
34. Daugaard G, Rossing N, Rorth M: Effects of cisplatin
on different measures of glomerular function in the human
kidney with special emphasis on high-dose. Cancer Chemother
Pharmacol 21:163-167, 1988
35. Collins JM, Zaharko DS, Dedrick RL, et al: Potential
roles for preclinical pharmacology in phase I clinical trials.
Cancer Treat Rep 70:73-80, 1986
36. van Hennik MB, van der Vijgh WFJ, Klein I, et al:
Comparative pharmacokinetics of cisplatin and three ana-
logues in mice and humans. Cancer Res 47:6297-6301, 1987
37. Edmonson JH, McCormack GM, Wieand HS, et al:
Comparison of cyclophosphamide + carboplatin vs cyclo +
cisplatin in stage III and IV ovarian carcinoma. Proc Am Soc
Clin Oncol 7:137, 1988 (abstr)
38. Wiltshaw E: Ovarian Trials at the Royal Marsden.
Cancer Treat Rev 1985; 12:67-71, (suppl)
39. Rodman JH, Abromowitch M, Sinkule JA, et al:
Clinical pharmacodynamics of continuous infusion teniposide:
Systemic exposure as a determinant of response in a phase 1
trial. J Clin Oncol 5:1007-1014, 1987