Disposition of recombinant human granulocyte colony-stimulating factor in children with severe chronic neutropenia

PEDIATRIC PHARMACOLOGY AND THERAPEUTICS

Disposition of recombinant human granulocyte colony-stimulating factor in children with severe chronic neutropenia Christine M. Kearns, PharmD, W i n f r e d C. W a n g , MD, N o r b e r t Stute, MD, J a m e s N. Ihle, PhD, a n d William E. Evans, PharmD From the Departments of Pharmaceutical Sciences, Hematology-Oncology, and Biochemistry, St. Jude Children's Research Hospital, Memphis, Tennessee, Center for Pediatric Pharmacokinetics and Therapeutics, and the Departments of Pediatrics and Clinical Pharmacy, University of Tennessee, Memphis

The disposition of recombinant human granulocyte colony-stimulating factor (G-CSF) was studied in 11 children with severe chronic neutropenia given 6 to 48 /~g G-CSF per kilogram subcutaneously. Serum concentrations of G-CSF were measured by bioassay. Peak serum G-CSF concentrations were proportional to d o s a g e and occurred 2 to 8 hours after subcutaneous administration. Nine of the eleven children had a significant increase in absolute neutrophil count (ANC). The median ANC in responding patients was 6.7 • t09/L on day 14 versus 0.17 • t09/L on day t of therapy (p <0.01). The G-CSF clearance increased a s A N C increased, and the relationship was well described by a sigmoid model. Maximal clearance a p p r o a c h e d 2 ml/min per kilogram at ANCs >17.0 • t09/L; minimal clearance was 0.29 ml/min per kilogram at ANCs of 0. The half-life of G-CSF was inversely related to ANC; mean half-life was 4.7 hours at ANCs of 0 but <2 hours at ANCs greater than 17.0 • t09/L. The two patients who failed to achieve a clinical response had no c h a n g e in G-CSF c l e a r a n c e or half-life, nor did they have an increase in ANC when G-CSF dosages were escalated to t8 or 48 #g/kg twice a day. These results indicate that G-CSF pharmacokinetics are directly influenced by ANC; higher serum concentrations, slower clearances, and longer half-lives are associated with low ANCs. (J PEDIATR1993;123:471-9) Granulocyte colony-stimulatingfactor (filgrastim) has been used successfully to treat congenital1-3 and cyclic4, 5 neutropenia and to ameliorate neutropenia resulting from myelosuppressive chemotherapy. 6-9 The dose and frequency of administration of G-CSF in these trials was determined empirically because little information was available on the disposition of G-CSF or on a concentration-effect relation-

Supported in part by Leukemia Program Project grant PO1 CA20180 and Cancer Center Core grant P30 CA21765, and by the American Lebanese Syrian Associated Charities. Submitted for publication March 11, 1993;accepted May 14, 1993. Reprint requests: W. E. Evans, PharmD, Pharmaceutical Department, St. Jude Children's Research Hospital, 332 N. Lauderdale, Memphis, TN 38101. Copyright | 1993 by Mosby-Year Book, Inc. 0022-3476/93/$1.00 + .10 9/25/48606

ship. Our primary goal was to determine the pharmacokinetics of G-CSF administered subcutaneously in pediatric patients receiving the hematopoietic growth factor for severe chronic neutropenia. Previous studies have demonstrated that G-CSF pharmacokinetics do not remain constant within patients being treated for chemotherapy-induced neutropenia and that G-CSF disposition is influenced by leukocyte count6' 7, 10 or duration of therapy. 6 Our second goal was to determine the effect of changes in absolute neutrophil count and the duration of G-CSF administration on the disposition of G-CSF in patients with severe chronic neutropenia. We assessed G-CSF pharmacokinetics in 1l pediatric and adolescent patients who were receiving recombinant human G-CSF subcutaneously for treatment of severe chronic neutropenia, obtaining a model of the relationship between G-CSF clearance and ANC response.

471

472

Kearns et al.

METHODS Patient enrollment. Pediatric patients receiving G-CSF according to a clinical protocol for the treatment of severe chronic neutropenia were eligible for evaluation in this pharmacokinetic study. Severe chronic neutropenia was defined as an A N C <0.5 X 109/L or, for patients with cyclic neutropenia, cycles with an A N C nadir <0.5 x 109/L on at least three occasions within a 6-month period. All patients underwent bone marrow biopsy and had serum evaluated for antineutrophil antibody before entry into the study. All studies were approved by the institutional review committee. Informed written consent was obtained from patients or their parents. Drug administration and sampling. All patients received recombinant human G - C S F provided by Amgen. The product (Neupogen) is a recombinant human protein expressed in Escherichia coli, has a molecular weight of 18,600 daltons, is nonglycosylated, and has a specific activity of approximately 2 X 108 units/mg protein. The G-CSF doses were administered at defined dosage levels as follows: level 1, 6 #g/kg daily; level 2, 6 ~g/kg twice a day; level 3, AIC ANC

c1/f C1MAX C1MIN CV G-CSF ke Vd

Akaike information criterion Absolute neutrophil count Clearance/fraction absorbed Maximum G-CSF clearance Minimum G-CSF clearance Coefficient of variation Granulocyte colony-stimulating factor Elimination rate constant Volume of distribution

12/~g/kg twice a day; level 4, 18/~g/kg twice a day; level 5, 24 ~g/kg twice a day; level 6, 36 ~g/kg twice a day; and level 7, 48 #g/kg twice a day. All patients received G-CSF by subcutaneous injection in the arm (deltoid muscle) or leg (thigh). All patients were sedentary for at least 2 hours after G-CSF injection. Patients with severe congenital neutropenia were started at dosage level 2; patients with cyclic neutropenia were started at dosage level 1. The clinical goal was to maintain an A N C consistently between 1.5 and 10.0 • 109/L. Patient response was evaluated after 14 to 28 days of therapy; if necessary, dosages were then adjusted. If the median A N C was <1.5 X 109/L during 14 days, the dosage was increased to the next level. If the A N C ranged between 1.5 and i0.0 • 109/L, the dosage was maintained. If the A N C was >25.0 • 109/L on three separate occasions or >10.0 • 109/L for 4 consecutive weeks, the dosage was decreased one level. Patient response was again evaluated after 14 to 28 days after dosage adjustments. Failure to maintain an A N C ___1.5 X 109/L, despite adjustments in the G-CSF dosage, classified a patient as a nonresponder. Samples for pharmacokinetic studies were obtained on

The Journal of Pediatrics September 1993

days 1, 4, and 14 of each dosage level. Samples were drawn before the dose was given, and at 0.25, 0.5, 1, 2, 4, 8, 12, and 24 hours after the dose was given. These sampling times were selected on the basis of a previous study of G-CSF in children with chemotherapy-induced neutropenia. 1~ One milliliter of blood was collected in tubes containing no anticoagulant. Samples were allowed to clot and then were centrifuged; the serum was stored at - 7 0 ~ C until assayed. Complete blood cell counts were obtained at least twice weekly and always on the day of pharmacokinetic sampling. Bioassay. Serum G-CSF was quantified by utilizing the cell proliferation assay of Shirafuji et al., 11 modified as previously described.l~ This bioassay of human G - C S F utilized a subclone of the murine leukemia cell line NFS-60 selected for its specific response to G-CSF and lack of cross-reactivity with other human cytokines, including interleukin-3, granulocyte-macrophage colony-stimulating factor, interleukin-6, erythropoietin, insulin-like growth factors 1 and 2, leukemia inhibitory factor, interleukin-1, and macrophage colony-stimulating factor (Dr. J. Ihle: unpublished data). All samples were assayed in duplicate. CelI proliferation, quantified by the incorporation of tritiated thymidine, was measured by scintillation counting. A sigmoidal model was fit to the disintegrations per minute versus G - C S F concentration dat a by using the A L L F I T software package (Biomedical Simulations Resource, University of Southern California, Los Angeles, Calif.). 12 Concentrations of G-CSF in serum samples and control samples were determined by analysis of simultaneous curve fittings of the samples and reference standards. The dilution yielding 50% of maximum proliferation (EC502) for samples was compared with the dilutions of reference standards of G-CSF (3 ng/ml and 30 ng/ml) giving 50% of maximum proliferation (EC501). The EC50z/EC501 ratio was then calculated, and G-CSF concentrations were calculated by multiplication of the ratio value by the reference standard concentration. The lower limit of assay detection was 0.03 ng/ml. The intraassay coefficient of variation was 12% at 3.9 ng G-CSF per milliliter, and the interassay CV was 18% when assessed at 24 ng G-CSF per milliliter. Complete patient courses (days l, 4, and 14) were assayed within a single-assay run to avoid interassay bias. Data analysis. A one-compartment model with zero-order absorption and first-order elimination was fit to the concentration-time data by using ADAPT II software (Biomedical).13 This pharmacokinetic model was chosen on the basis of previous G-CSF studies in our laboratory) ~ A Bayesian algorithm with published parameters as "priors ''6-1~ was used to fit this model to the data, permitting pharmacokinetic estimation with a clinically feasible number of blood samples. The prior values for elimination rate constant (k~, 0.14 H - I ; CV, 50%) and apparent volume of

The Journal o f Pediatrics Volume 123, Number 3

Kearns et al.

400 3

350

==

300

473

|

t--

.~

250

c

200

0 c 0

150

0 "r

100

(9

50

o,

|

|

?

0 6 qd

6 bid

12 bid

|

18 bid

24 bid

36 bid

48 bid

G-CSF Dosage (mcg/kg/~lose)

Fig. I. Measured peak serum concentrations of recombinant human G-CSF administered subcutaneously after day 1, dose 1, of indicated dosage levels. Patients responding to therapy are indicated by a clear square (E]), response failures as a clear circle (9 Numbers within symbols correspond to individual patient numbers in Tables I and lI. T a b l e I. Patient characteristics, diagnosis, and dosing schedules

Patient No.

Age (yr)

Sex

Diagnosis

Pretheropy G-CSF (ng/ml)

1 2 3 4 5 6 7 8 9 10 11

12 15 18 5 18 1 1 1 5 12 10

F F F F M M M F M F F

CN CN CN CN CN CN CN CN Hyper IgM Cyclic N Cyclic N

0.06 0.73 Undetectable* 0.11 0.!2 0.09 NA 0.05 0.84 0.79 0.03

Dosage Initial 6 gg/kg 6 gg/kg 6 gg/kg 6 #g/kg 6 #g/kg 6/zg/kg 24 ug/kg 6 #g/kg 6 ~g/kg 6 #g/kg 6 ~g/kg

b.i.d. b.i.d. b.i.d. b.i.d. b.i.d. b.i.d. b.i.d.t b.i.d. b.i.d. q.d. q.d.

Final

Clinical response

6 gg/kg b.i.d. 6 ~zg/kg b.i.d. 6 #g/kg q.d. 6 #g/kg b.i.d. 6 ~g/kg q.d. 18 ttg/kg b.i.d. 48 ug/kg b.i.d. 6 gg/kg b.i.d. 6 #g/kg b.i.d. 6 #g/kg q.d. 6 jzg/kg q.d.

' Yes Yes Yes Yes Yes No No Yes Yes Yes Yes

CN, Congenital neutropenia; NA, not available; Cyclic N, cyclic neutropenia. *<0.03 ng/rnl. tG-CSF therapy initiated before study.

distribution (Vd 100 ml(kg; c v , 80%) used in the model were extracted from reports of patients with A N C values ranging from 0 to 16.0 • I0 9/L. The mode ! was formulated as a closed-form solution parameterized for k~ and apparent Vd. Apparent systemic Clearance, elimination half-life and a r e a under the concentration-time curve were calculated as secondary parameters. Peak serum concentration was defined as the highest measured value at each dosage level. T h e Zero-order absorption rate constant was estimated for each dosage on the basis of the dose administered and the time when absorption occurred. If G - C S F was measured in predose samples (time = 0), the amount was accounted for by Specifying this concentration as an initial condition in the model.

First-, second-, and third-degree polynomials, as well as a sigmoidal model, were fit to the G - C S F clearance versus A N C data. All patients (responders and nonresponders) and all data pairs of G-CS F Clearance and A N C values were included. The A L L F I T software package tz was used to fit a sigmoidal curve to the G - C S F clearance and A N C data, as described by the following equation: C1

C1MIN

--(- = C1MAX-- 1 + (ANC/ANC5o) s + C1M~N where C l / f equals the apparent clearance of subcutaneously administered G-CSF, CIMAx is the maximum G - C S F clearance, C1Mm is the minimal G - C S F clearance, ANCso is tbe A N C at which the G - C S F clearance is 50% of the

474

Kearns et al.

The Journal o f Pediatrics September 1993

T a b l e II, Peak G - C S F concentrations, a p p a r e n t clearance, elimination half-life, and A N C on days 1, 4, a n d 14 of treatment Cmax ( n g / m l ) 1" Dosage = Patient 3 5 10t 11# Dosage = Patient it 2t 3# 4# 5# 6# 8# 9# Dosage = Patient 6 Dosage = Patient 6 Dosage = Patient 7# Dosage = Patient 7 Dosage = Patient 7

4

6/zg/kg q.d. No. 12.6 17.9 7.3 8.4 35.3 10.3 6/zg/kg b.i.d. No. 39.2 32.3 45.0 51.1 42.1 -15.6 36.5 14.9 7.7 16.6 24.6 32.9 -50.5 37.1 12 #g/kg q.d. No. 40.1 31.3 18 izg/kg b.i.d. No. 72.5 44.9 24 izg/kg b.i.d. No. 178.0 135.0 36 ~g/kg b.i.d. No. 197 229 48 lzg/kg b.i.d. No. 380 330 -

-

-

-

Cl/f ( m l / m i n / k g )

ANC (Xt09IL)

tY2 (hr)

14.

t

4

t4.

t

4

t4

t

4

14,

-9.7 3.9 24.5

0.88 1.00 1.03 0.35

0.92

0.54

-1.07 1.96 0.77

1.9 1.4 1.8 3.3

1.0 --3.3

-1.7 1.0 2.4

16.7 11.2 1.7 1.7

9.7 4.7 3.4 0.2

3.0 4.0 19.5 6.7

9.7 48.2 6.6 -2.8 14.8 13.9 35.9

0.37 0.13 0.26 0.42 0.44 0.43 0.47 0.19

0.34 0.19 -0.33 1.02 0.39 -0.23

1.57 0.25 1.45 -1.87 0.61 0.74 0.29

3.8 7.1 4.3 3.7 1.1 3.2 3.2 6.3

4.1 5.1 -3.5 1.7 3.8 -4.3

1.2 3.3 1.3 -0.9 2.9 1.2 4.6

0.1 0.~ 0.3 0.1 0.2 0.1 0.0 0.6

0.0 1.7 0.05 0.0 16.4 0.2 -0.04

5.2 5.8 10.1 6.2 17.7 1.1 4.1 16.9

56.1

0.46

0.37

0.44

3.7

3.5

3.2

0.0

0.3

0.6

78.6

0.44

0.64

0.25

3.7

2.9

6.3

1.6

14.4

0.1

--

0.19

0.27

--

6.9

4.1

0.0

0.0

--

0.27

0.30

--

3.6

3.3

0.1

0.4

--

0.26

0.26

--

5.8

3.9

0.0

0.0

-

-

-

-

Cmax, Peak serum concentration t89 half-life; tx, treatment; --, data not available.

*Measurements listed in this table were obtained on days 1, 4, and 14 of treatment. tEntry dosage regimen. m a x i m a l clearance, a n d S is the slope factor for the sigmoid curve. 12This model was used to estimate the A N C s 0 a n d the slope of the sigmoid curve for the study population, with C1MIN and C1MAX set as fixed parameters. T h e c1Mi N was fixed at 0.293 m l / m i n per kilogram, the m e a n value for the clearance of G - C S F in patients with an A N C of 0 (10 observations in 7 patients). Because our model was confined to A N C values < 2 0 X 109/L, C1MAX was defined as the m e a n c l e a r a n c e (i.e., 1.92 m l / m i n per kilogram) in the two patients with the highest A N C values ( 17.7 and 19.5 X 109/ L). Fixing C1MIN a n d C1MAX improved the statistical measures of goodness of fit for the model (see Statistical Analysis section, below). Statistical analysis. Differences in the pharmacokinetic p a r a m e t e r s and A N C values between days 1 and 4 a n d between days 1 and 14 were analyzed by the Wilcoxon signed r a n k test. N o n p a r a m e t r i c statistical tests were used because

the data (e.g., p h a r m a c o k i n e t i c parameters, A N C ) were not normally distributed. Comparisons of peak concentration a n d clearance in patients receiving a dosage of 6 # g / k g twice a day were performed for patients with data on b o t h days of study (day l vs day 4, n -- 6; day 1 vs day 14, n = 7). T h e Wilcoxon signed r a n k test was also used to compare A N C values and clearances on days 1 and 14 of therapy in the subgroup of patients classified as responders. T h e test was performed for patients with data on both days of study ( A N C , n = 9; clearance, n = 8). Appropriateness o f the models of G - C S F clearance versus A N C was assessed by the Akaike i n f o r m a t i o n criterion. 14 M i n i m i z a t i o n of A I C , a function incorporating the n u m b e r of data points, the sum of squared residuals, and the n u m b e r of model parameters, was used to select the optimal model. For the sigmoidal model, goodness of fit of the model with and without fixed p a r a m e t e r s was assessed by exam-

The Journal o f Pediatrics Volume 123, Number 3

Kearns et al.

475

100

30 -~

10

-

C

c ._o

3

c

]

0 c 0

0

0.3 0.1

-

0.03 0

I

I

I

I

I

I

2

4

6

8

10

12

I

Time (hours)

Fig.

2. Concentration-versus-time plots for day l, dose 1, of therapy in patients receiving recombinant human G-CSF, 6 #g/kg twice a day, subcutaneously. Symbols represent measured serum concentrations; lines are model simulations based on each patient's pharmacokinetic parameter estimates.

ination of the residuals for equal distribution around zero and lack of heteroscedasticity, minimization of the mean square error, and minimization of the sum of squares of the data and model estimates. RESULTS Eleven patients with severe chronic neutropenia were studied (congenital neutropenia, n = 8; cyclic neutropenia, n = 2; hyperimmunoglobulin M syndrome, n = 1). There were 7 female and 4 male patients; the median age was 10 years (Table I). Endogenous G-CSF serum concentrations measured before the first dose of recombinant G - C S F 9 ~-~ 1Tanged from undetectable (<0.03 ng/ml) to 0.84 ng/ml, Wi~h a median of 0.10 ng/ml. After subcutaneous administration of G-CSF, measured peak concentrations occurred between 2 and 8 hours after the dose. There was no apparent difference in G - C S F disposition when it was administered subcutaneously in the arm versus the leg. Measured peak G - C S F concentrations correlated directly with the dose administered (Fig. 1). In patients receiving an initial dose of 6 #g/kg, peak concentrations on day 1 of therapy ranged from 8.4 to 50.5 ng/ml (mean 30.1 ng/ml). Measurable serum concentrations were maintained throughout the entire 12-hour dosing interval in patients receiving twice-daily dosing. Only one patient (No. 5) was at >95th percentile for body weight; he had the lowest peak concentration on day 1 at both dosage levels studied (i.e., 6 mg/kg daily or twice a day) (Table II). Two patients received dosage escalations because of lack of clini-

cal response (Table II). In patient 6, dosage increments to 12 #g/kg twice a day and 18 #g/kg twice a day resulted in peak concentrations of 40.1 and 72.5 ng/ml, respectively. In patient 7, a dosage of 24 #g/kg twice a day resulted in a measured peak concentration of 178 ng/ml; a dosage of 36 #g/kg twice a day resulted in a peak concentration of 197 ng/ml; and the highest dosage administered, 48 #g/kg twice a day, yielded a peak concentration of 380 ng/ml. Pharmacokinetic studies were performed on days 1, 4, and 14 of the entry dosage level and again if the dosage was adjusted. Mean values of the measured peak serum concentrations for the eight patients receiving a dosage of 6 #g/kg twice a day were 32.1 ng/ml on day 1 (n = 8), 31.6 ng/ml on day 4 (n = 6), and 18.8 ng/ml on day 14 (n = 7) (Table lI). Measured peak serum concentrations for patients with data on both days 1 and 14 differed significantly (p = 0.05; Wilcoxon signed rank test). For the eight patients, Fig. 2 depicts the concentration-time profile for day 1 of therapy with an initial dosage of 6 #g/kg twice a day. The range of times of peak G-CSF concentrations and relative uniformity in elimination are evident. Median G-CSF clearance after subcutaneous dosing on day 1 of therapy in patients receiving a dosage of 6 #g/kg twice a day was 0.40 ml/min per kilogram and ranged from 0.13 to 0.47 ml/min per kilogram. On day 4 of therapy, the median clearance was 0.34 ml/min per kilogram, ranging from 0.19 to 1.02 ml/min per kilogram, and was not statistically different from day 1. By day 14 the median clearance of G-CSF was 0.74 ml/min per kilogram, with a range of

476

Kearns et al.

The Journal of Pediatrics September 1993

2.0

[] []

1.5 Q

1.0

In an analysis of the changes in clearance with respect to ANC, first-, second-, and third-degree polynomial models yielded fits inferior to the sigmoidal model, as assessed by the AIC. The model fit of the relationship between G-CSF clearance rates and ANC is depicted in Fig. 4. The baseline clearance of G-CSF (0.29 m l / m i n per kilogram) remained relatively constant through ANC values UP to 1.0 X 109/L. As predicted by this model, at ANCs between 1.0 x 109/L and 100 X 109/L, G-CSF clearance increased in an approximately linear relationship to the logl0 of ANC.

U_

o,

DISCUSSION

(..9 0.5

I

I

I

Day 1

Day 4

Day 14

Fig. 3. Clearance of recombinant human G-CSF administered subcutaneouslyon days 1, 4, and 14 of initial course of therapy in patients classified as responders to therapy. Clear diamonds ((?) indicate a dosage of 6 #g/kg daily, clear squares (t-l) a dosage of 6 #g/kg twice a day. Numbers within symbols refer to individual patient numbers (Tables I and II). Clearance day 1 versus day 14 differed significantly (p = 0.01, Wilcoxon; n = 8)./. 0.25 to 1.87 ml/mi n per kilogram. Clearance was significantly different on day 14 versus day 1 of therapy in patients with data for both days (Wilcoxon: n = 7; p = 0.02). In this patient group, terminal ha!f-life appeared to decrease from a median of 3.8 hours on day 1 to a median of 1.3 hours on day 14. The median Vd apparent was 107 ml/kg on day 1 and 153 ml/kg on day 14 (Wilcoxon: p = 0.052). Of the 11 patients, 9 tiad increases in their ANCs and were classified as responders to therapy (Table I). The ANC values before the first dose of G-CSF therapy ranged on day 1 from 0 to 1.7 x 109/L (median 0.2 X 109/L), on day 4 from 0 to 16.3 X 109/L (median 0.1 • 109/L), and by day 14 from 4.1 to 19.5 X 109/L (median 6.7 x 109/L). The ANC for day 1 versus day 14 in these nine patients differed significantly (Wilcoxon: p <0.01). The median apparent clearance of G-CSF was 0.37 ml/min per kilogram on day 1, increasing to 1.11 ml/min per kilogram by day 14 of therapy (Fig. 3). In patients with studies on both days 1 and 14 (n = 8), the differences in clearance were statistically significant (wilcoxon: p = 0.01). However, in the two patients who did not have a clinical response (patients 6 and 7), the clearance of G-CSF did not change appreciably from day 1 to day 14 of therapy, nor were there changes in clearance from baseline values even after dosage escalations.

Our study characterized the disposition of recombinant human G-CSF in pediatric patients with severe Chronic neutropenia. Endogenous G-CSF serum concentrations were measurable before therapy in 9 of 10 patients (pretreatment sample unavailable in one patient), consistent with previous reports of elevated circulating G-CSF concentrations in patients with neutropenia. 15-17 The concentration of endogenous G-CSF measured before therapy was not a predictive factor for response to G-CSF therapy in this study (Table I). After subcutaneous dosing, measurable serum concentrations of G-CSF were maintained during at least a 12-hour period in patients receiving doses > 6 ~zg/kg. The peak concentrations achieved by a single dose of 6 #g/kg were 70- to 500-fold greater than the median endogenous circulating G-CSF concentration before G-CSF therapy. Although the precise mechanisms responsible for the sevenfold range in peak concentrations are unknown, higher peak concentrations occurred in patients with low ANCs and vice versa. This observation is consistent with neutrophils acting as a presystemic clearance mechanism for G-CSF. Between days 1 and 4 of therapy, the disposition of recombinant human G-CSF was not statistically different, nor were there significant changes in ANC values. All patients who had an objective clinical response did so by day 14 of their initial course of therapy; this was reflected by a statistically significant increase in ANC values and a concomitant increase in G-CSF clearance. As would be expected, peak concentrations of G-CSF declined twofold to threefold after an ANC response, reflecting the twofold to threefold increase in G-CSF clearance. Reports in adult and pediatric populations receiving recombinant G-CSF for chemotherapy-induced neutropenia have documented similar changes in G-CSF plasma or serum concentrations after neutrophil responses.6s, s0 Conversely, in patients classified as nonresponders to therapy, clearance did not change appreciably from baseline values, despite prolonged periods

The Journal o f Pediatrics Volume 123, Number 3

Kearns et al.

477

2.5

2.0

[] []

i.-

~

c

[] 1,5 (D o

~

,/

[] D ~ D

_.e s 1.0 LL 03

[]

o, 0.5

13] E1

0.0

I '

o/

2

[] ,001

i

i

i

iiiiii

i i

J I IIIIIIi

0.01

I

0.1

I I fllll

I

1,0

I

I I IIIII

I

10

I

[ I Illl

1

100

ANC (X 109/L)

Fig. 4. Re•ati•nshipbetweenANCandc•earance•frec•mbinanthumanG-CSFina••patients.WhenC•MAxandC•M•N were fixed at 1.92 ml/min per kilogram and 0.29 ml/min per kilogram, respectively,the estimated ANC at which G-CSF clearance was 50% of maximum (ANCs0) was 10.5 • 109/L (SEM 2.4; CV 22.5%). The slope of the sigmoid curve "S" was +0.93 (SEM 0.24; CV 25.5%).

of therapy and dosage escalations. These data suggest that the changes in G-CSF disposition were related to the number of circulating neutrophils and that duration of therapy, per se, had little or no effect on G-CSF pharmacokinetics. The half-life of G-CSF appeared to decrease as neutrophil counts and G-CSF clearance increased. The frequent dosing of G-CSF in our patient population, usually every 12 hours, limited our ability to measure serum G-CSF concentrations through several half-lives. Nevertheless, the G-CSF half-life became shorter in all patients whose neutrophil count increased during G-CSF therapy. Our assessment of half-fife may have been limited by the clinical context within which these studies were conducted, but the observed change in half-life was consistent within our population and in agreement with previous studies. 6, lo Because mature neutrophils bind G-CSF and can potentially act as a clearance mechanism, we constructed a model of the clearance of G-CSF as predicted by the ANC. A sigmold model with fixed limits of maximal and minimal clearance values was fit to the data; the slope of the curve, representing changes in clearance as a function of the ANC, and the ANC at which G-CSF clearance was 50% of maximum (ANC50) were estimated. The model predicted a relatively constant G-CSF clearance at ANC values <1.0 X 109/L, consistent with clearances measured in our patients. This clearance value, approximately 0.29 ml/min per kilo-

gram, likely represents the intrinsic clearance of G-CSF in the absence of clearance by circulating neutrophils, by mechanisms that may include metabolism of the protein and renal elimination. As the A N C increased to greater than 1.0 X 109/L, G-CSF clearance increased substantially, in a linear relationship to the loglo of the ANC: Thus clearance more than trebled as the A N C rose from 1.0 to 20 X 109/L. In our previously published study of children with chemotherapy-induced neutropenia, ~~the C1MAXwas 2.5 ml/min per kilogram in children with high ANCs (> l 5 x 109/L). Using the data from the current study, we constructed a model by using 2.5 ml/min per kilogram instead of 1.92 ml/min per kilogram as the fixed C1MAX. This analysis yielded a nearly identical slope value "S" and an estimate of ANCs0 of 20.6 X 109/L (SEM 5.7). Thus with data from these two studies it appears that G-CSF clearance reaches a half-maximal rate at ANCs of 10.0 to 20.0 X 109/L. One limiting feature of our model is that ANCs were confined to values between 0 and 19.5 x 109/L. It is possible, for example, that G-CSF clearance would continue to increase in a linear fashion at ANCs greater than 100 X 109/L, the ANC at which our model predicts maximal G-CSF clearance. Although additional information on the clearance of G-CSF at ANC values _>20 • 109/L would be desirable from the standpoint of validating the model, our model does

478

Kearns et al.

encompass the range of normal A N C values and the range of A N C values observed in our population from diagnosis through the desired response. Clearance determined after subcutaneous administration of a drug is affected by both absorption into the systemic circulation from the injection site and removal of drug from serum. Thus either reduced subcutaneous absorption or enhanced removal from the serum would result in higher estimates of G - C S F clearance after subcutaneous administration. However, the decrease in elimination half-life seen with increasing A N C in our present study suggests that changes in clearance were due primarily to changes in systemic elimination and not changes in absorption. Moreover, the shorter half-life was observed despite an increase in Vd, which would be expected to prolong the half-life (0/2) in the absence of a change in clearance (C1/f): tl/2 = 0.693 X Vd/

cl/f. Administration of recombinant G - C S F produced good neutrophil responses in eight of nine patients with endogenous circulating G - C S F concentrations. The fact that neutrophil responses were induced by supraphysiologic concentrations of G - C S F is consifitent with a hypothesis of relative inadequacy of endogenous G - C S F production, perhaps because of decreased receptor affinity, decreased receptor avidity, is or impaired but not absent signal transduction pathways. 19 Only in patient 3, who had no measurable circulating endogenous G - C S F before therapy, was the cause of neutropenia consistent with the absence of endogenous G - C S F production. The two patients who responded poorly to G - C S F therapy (patients 6 and 7), even at increased dosages, may represent patients with aberrant or even absent receptors or signal transduction pathways. After more than 1 year of therapy, patient 2, initially a responding patient, became less responsive to therapy despite dosage increases. The exact cause of this change has yet to be determined; antineutrophil antibody and anti-G-CSF antibody titers have remained undetectable. On the basis of measured G - C S F plasma concentrations, the lack of response in these patients was not related to ineffective absorption or enhanced clearance of G-CSF. In our study, 6 ~g of recombinant human G - C S F per kilogram, given subcutaneously once or twice daily, was sufficient to produce responses in 9 of 11 patients with severe chronic neutropenia; further dosage escalations were not immediately beneficial in nonresponders, despite proportionally higher G - C S F serum concentrations. More data are needed to determine, with certainty, the G - C S F serum concentration beyond which escalation of treatment is ineffective. Characterizing the relationship between concentration of G - C S F and neutrophil response should help clinicians provide optimal therapy, with minimal total drug exposure,

The Journal of Pediatrics September 1993

to patients who respond to G-CSF, and may identify patients unlikely to respond to G-CSF, indicating that alternative therapeutic modalities may be needed. We thank H. Clariette and S. Wooten for their technical assistance with the bioassay, K. Silverstein, MS, Yuri Yanishevski, and Nancy Kornegay for software support, and B. Arnold, RN, for assistance with the clinical data. We also thank Dr. J.-S. Lia for statistical advice, and Drs. William Crom and John Rodman for consultation with biomedical modeling.

REFERENCES

1. Bonilla MA, Gillio AP, Ruggeiro M, et al. Effects of recombinant human granulocyte colony-stimulating factor on neutropenia in patients with congenital agranulocytosis. N Engl J Med 1989;320:1574-80. 2. Welte KA, Zeidler C, Reiter A, et al. Differential effects of granulocyte-macrophage colony-stimulating factor and granulocyte colony-stimulating factor in children with severe congenital neutropenia. Blood 1990;75:1056-63. 3. Jabowski AA, Souza L, Kelly F, et al. Effects of human granulocyte colony-stimulating factor in a patient with idiopathic neutropenia. N Engl J Med 1989;320:38-42. 4. Hammond WP, Price TH, Souza LM, Dale DC. Treatment of cyclic neutropenia with granulocyte colony-stimulating factor. N Engl J Med 1989;320:1306-11. 5. Migliaccio AR, Migliaccio G, Dale DC, Hammond WP. Hematopoietic progenitors in cyclic neutropenia: effect of granulocyte colony-stimulating factor in vivo. Blood 1990;75: 1951-9. 6. Layton JE, Hockman H, Sheridan WP, Morstyn G. Evidence for a novel in vivo control mechanism of granulopoieses: mature cell-related control of a regulatory growth factor. Blood 1989;74:1303-7. 7. Sheridan WP, Morstyn G, Wolf M, et al. Granulocyte colonystimulating factor and neutrophil recovery after high-dose chemotherapy and autologous bone marrow transplantation. Lancet 1989;2:891-5. 8. Morstyn G, Campbell L, Lieschke G, et al. Treatment of chemotherapy-induced neutropenia by subcutaneously administered granulocyte colony-stimulating factor with optimization of dose and duration of therapy. J Clin Oncol 1989;10:155462. 9. Morstyn G, Campbell L, Souza LM, et al. Effect of granulocyte colony-stimulating factor on neutropenia induced by cytotoxic chemotherapy. Lancet 1988;1:667-72. 10. Stute N' Santana VM' R~ JH' Schell MJ' Ihle JN' Evans WE. Pharmacokinetics of subcutaneous recombinant human granulocyte colony-stimulating factor in children. Blood 1992;79:2849-54. 11. Shirafuji N, Asano S, Matsuda S, Watari K, Takaku F, Nagata S. A new bioassay for human granulocyte colony-stimulating factor (hG-CSF) using murine myeloblastic NFS-60 cells as targets and estimation of its levels in sera from normal healthy persons and patients with infections and hematological disorders. Exp Hematol 1989;17:116-9. 12. DeLean A, Munson P J, Rodbard D. Simultaneous analysis of families of sigmoidal curves: application to bioassay, radioli-

The Journal o f Pediatrics Volume 123, Number 3

13.

14.

15.

16.

gand assay, and physiological dose-response curves. Am J Physiol 1978;235:E97-E102. D'Argenio DZ, Schumitzky A. ADAPT 1I user's guide. Los Angeles: Biomedical Simulations Resource, University of Southern California, 1990. Yamaoka K, Nakagawa T, Uno T. Application of Akaike's information criterion (AIC) in the evaluation of linear pharmacokinetic equations. J Pharmacokinet Biopharm 1978;6: 165-75. Mempel K, Pietsch T, Menzel T, Zeidler C, Welte K. Increased serum levels of granulocyte colony-stimulating factor in patients with severe congenital neutropenia. Blood 1991; 77:1919-22. Watari K, Asano S, Shirafuji N, et al. Serum granulocyte col-

Freeman et al.

479

ony-stimulating factor levels in healthy volunteers and patients with various disorders as estimated by enzyme immunoassay. Blood 1989;73:117-22. 17. Yujiri T, Shinohara K, Kurimoto F. Fluctuations in serum cytokine levels in the patient with cyclic neutropenia. Am J Hematol 1992;39:144-5. 18. Abboud M, Firpo M, Laver J, et al. Studies of G-CSF production and response in congenital neutropenia [Abstract]. Exp Hematol 1989;17:671. 19. Kyas U, Pietsch T, Welte K. Expression of receptors for granulocyte colony-stimulating factor on neutrophils from patients with severe congenital neutropenia and cyclic neutropenia. Blood 1992;79:1144-7.

Clinical and laboratory observations Intravenously administered immune globulin for the treatment of infection-associated hemophagocytic syndrome Bridget Freeman, MD, M o b e e n H. Rathore, MD, Emad Salman, MD, M i c h a e l J. Joyce, MD, PhD, a n d Paul Pitel, MD From the Departments of Pediatrics, University of Florida Health Science Center and Nemours Children's Clinic, Jacksonville, Florida

Infection-associated h e m o p h a g o c y t i c syndrome is an unusual disease with a high mortality rate. A variety of treatment modalities have been used with limited success. We report three patients with infection-associated h e m o p h a g o cytic syndrome successfully treated with intravenously administered immune globulin. (J PEDIATR1993;123:479-81) Infection-associated hemophagocytic syndrome is characterized by a benign proliferation of histiocytes throughout the reticuloendothelial system in association with a systemic infection, often viral. The immune mechanism is not clear,

Submitted for publication Jan. 13, 1993; accepted April 22, 1993. Reprint requests: Mobeen H. Rathore, MD, 653-1 W. 8th St., De~partment of Pediatrics, Jacksonville, FL 32209. Copyright 9 1993 by Mosby-Year Book, Inc. 0022-3476/93/$1.00 + .10 9/26/48132

but if untreated, this disease has a high mortality rate.l Numerous treatment modalities have been used, with limited success. We report three cases of I A H S in which intravenously administered immune globulin was used successfully. CASE REPORTS Patient 1. A 7-year-old black girl had a 1 week history of fever (up to 39.4 ~ C) and lethargy. Her examination on admission to the hospital showed a temperature of 38.3 ~ C. She was lethargic but responded appropriately to questioning and had no focal neurologic signs. There was no significant lymphadenopathy or bepatosple-