Pediatr Blood Cancer 2006;46:723–727

REVIEWPotential Clinical Impact of Taking Multiple Blood Samples

for Research Studies in Paediatric Oncology:How Much Do We Really Know?

Michael Cole, BSc, MSc, FSS, CStat,1 Alan V. Boddy, BSc(Hons), PhD,1 Pamela Kearns, BSc(Hons), PhD, MBChB, MRCP, MRCPCH,3

Kok H. Teh, MBBS,2 Lisa Price, SRN, RSCN,2 Annie Parry, RGN, RSCN,2 Andrew D.J. Pearson, MD, FRCP, FRCPCH,2

and Gareth J. Veal, BSc(Hons), PhD1* on behalf of the UKCCSG Pharmacology group

INTRODUCTION

In undertaking research studies in patients with cancer, it

is essential to consider every aspect of a study in terms of the

ethics of the proposed research and the likely social,

psychological and clinical impact on the patients. For

example, for many patients with cancer it is often the case

that quality as well as duration of life will be an important

consideration. While patients are often willing to experience

a certain amount of discomfort and inconvenience in order to

participate in a research study, this is balanced by a degree of

trust, placed in the investigators, that such participation will

not result in an increased risk to patient health. These issues

are even more acute when considering clinical research

studies in children with cancer, due to the vulnerable nature

of this patient group.

For clinical studies investigating new treatments or drug

combinations in the quest to improve survive in paediatric

oncology, there may be clinical benefit to those children

participating. However, there are many research studies

where the benefit of participating in a studymay be less clear-

cut. A good example of this would be where a clinical

pharmacology study is being carried out in order to learn

about pharmacokinetics in a paediatric patient population.

The justification for such research is that it will be of use in

the optimisation of treatment for similar patients receiving a

particular drug or chemotherapy regimen in the future.

However, the data will not directly benefit the children

actually participating in the study. In this scenario, it is

clearly essential to minimise any risk from participating in

these studies.

Clinical Pharmacology Studies

The study of the clinical safety and efficacy of chemo-

therapeutic agents in children is essential because both can

differ considerably from that observed in adults. As these

studies require ex vivo measurements in plasma, blood

sampling is unavoidable with the currently available

technologies. There are many examples of drugs where data

obtained from early clinical studies in adults do not reflect

pharmacokinetic and/or pharmacodynamic relationships in

paediatric studies [1,2]. In addition, variationsmay be seen in

the way that drugs are handled among paediatric subpopula-

tions such as newborns, infants and adolescents [3,4]. For

example, the continued development of hepatic and renal

function in very young children may result in variable drug

toxicity profiles due to alteredmetabolism or elimination [5–

7]. For many of the chemotherapeutic drugs currently used in

paediatric oncology, specific pharmacological data are

lacking as these data were not required when the drugs were

licensed for use in adult patients.

Pharmacological studies of anti-cancer agents in children areessential to determine their clinical safety and efficacy, both of whichcan differ considerably from that observed in adults. However, thepotential clinical impact of taking blood samples, in addition to thoserequired for standard clinical practice is commonly a concern forboth medical and allied staff and parents. Frequently quoted ‘safelimits’ of 3%–5% of total blood volume taken on any one study dayare not based on published data and may not be acceptable for allpatients. This article reviews some of the reasons why clinical

pharmacology data for anti-cancer drugs is often lacking in apaediatric patient population, summarises data from a retrospectivestudy investigating the potential impact of repeated blood samplingfor research purposes and discusses how this issue may be moresystematically addressed in future studies. Research involvingchildren with cancer should be limited to those studies addressingkey scientific questions and should be designed to limit both thenumber and volume of blood samples required. Pediatr BloodCancer 2006;46:723–727. � 2005 Wiley-Liss, Inc.

Key words: blood sampling; cancer; children; clinical research; ethics

� 2005 Wiley-Liss, Inc.DOI 10.1002/pbc.20463

——————1Northern Institute for Cancer Research, University of Newcastle upon

Tyne, Newcastle upon Tyne, United Kingdom; 2Department of Child

Health, University of Newcastle upon Tyne, Newcastle upon Tyne,

United Kingdom; 3Department of Clinical Sciences, Institute of Child

Life and Health, University of Bristol, Bristol, United Kingdom

*Correspondence to: Gareth J. Veal, Northern Institute for Cancer

Research, Paul O’Gorman Building, Medical school, Framlington

Place, University of Newcastle upon Tyne, Newcastle upon Tyne, NE2

4HH, UK. E-mail: G.J.Veal@newcastle.ac.uk

Received 11 February 2005; Accepted 18 April 2005

In addition to commercial considerations, pharmacologi-

cal data may be lacking in children due to problems with the

recruitment of sufficient patient numbers for a meaningful

study. Alternatively, there may be difficulties in obtaining the

required number of samples or blood volumes, particularly in

smaller children [8]. Where comprehensive clinical pharma-

cology studies have been conducted, significant changes in

clinical practice have resulted for drugs such as methotrexate

and carboplatin [9–12]. It is therefore essential that clinical

pharmacology studies are undertaken both to optimise the

treatment of children with cancer and to facilitate the

development of novel agents.

Potential Risks—How MuchDo We Really Know?

A common concern for nurses, clinicians and the parents

of children who are approached to participate in clinical

pharmacology research studies is the potential clinical

impact of blood sampling on the patient. A figure of 5% of

total blood volume (calculated from body weight and surface

area) is often quoted as an acceptable, safe limit for blood

sampling on any one day of a study [13]. Although, in

practice, most studies actually requiremuch smaller volumes

of blood for analysis, there is no scientific basis for this limit.

With haematological toxicity representing a frequent side-

effect of many anti-cancer drugs used in paediatric oncology,

concerns over thewithdrawal of additional blood samples for

pharmacokinetic research studies, and the potential link with

the requirement of patients to undergo blood transfusions, are

understandable. In addition, there are possible risks asso-

ciated with the frequency with which a central venous

catheter (CVC), or similar central line, may be accessed,

particularly relating to the risk of infection. Whilst this may

represent a small percentage of the total number of times that

a line is accessed for routine clinical samples to be drawn, it

may be considered an increased risk of CVC infection. From

an ethical point of view it is important that the potential

impact of repeated blood sampling for research purposes is

understood, in order to give some confidence in the clinical

relevance of the current ‘best practice’.

Retrospective Study

In order to address this issue, we selected a recently

published clinical pharmacology study involving children

with cancer for which we had been responsible for

patient recruitment and study participation. This study

involved measurement of drug-DNA adduct levels in

peripheral blood leucocytes obtained from children receiving

cisplatin. A comparison was made between DNA adducts

formed when pre-treatment blood samples were incubated

with cisplatin in vitro, and those observed in patients treated

with this drug in the clinic [14]. In order to perform the invitro

investigations, a relatively large pre-treatment sample was

required, in addition to blood samples for cisplatin

pharmacokinetics. In this study, the total volumes of blood

taken per patient were often near to the 5% quoted ‘safe

limits’. This therefore represented an ideal choice for a

retrospective study as it maximised the likelihood of a

negative clinical impact being observed. In addition, this was

a single centre study, so access to the relevant clinical data

was possible.

In order to investigate whether or not participation in this

pharmacology research study had any measurable impact on

the clinical welfare of the participants, a retrospective

analysis of clinical data from these patients was performed.

Clinical data were obtained from eleven patients who

participated in the research study. Eight of these patients

had been diagnosed with osteosarcoma, two with neuroblas-

toma and one patient with hepatoblastoma. Each patient

received five or six courses of standard chemotherapy.

Additional blood samples were taken over a 3-day period for

research purposes on one of these courses of treatment. Six

samples (3ml each) were taken for pharmacokinetic analysis

and a further four samples (10 ml each) obtained for

measurement of drug-DNA adduct levels. It should also be

noted that a 5 ml discard volumewould also be taken prior to

the withdrawal of these blood samples from the patient’s

central line. This resulted in total blood volumes of between

50 and 90 ml being taken over the 3-day study period.

Clinical data including haemoglobin level, white cell and

platelet count were collected following courses where only

clinically-indicated blood samples were taken, and those

courses where additional samples were taken for research

purposes. With this design, each patient acts as their own

control. Additional data obtained included patient weight,

age and sex, together with information on the frequency and

timing of blood transfusions during each course of

chemotherapy and the volume of blood taken during the

research study.

Retrospective Data Analysis

Despite the fact that statistical analysis of data from

this study involved the retrospective amalgamation of a

relatively large amount of clinical data, including haemo-

globin, white cell and platelet counts from a total of

64 courses of chemotherapy, detailed clinical data were

available in all cases. Analysis of haematological data

showed that there was a mean additional drop in nadir

haemoglobin levels of 1.5 g/dl when blood samples were

taken for research purposes, after adjusting for between-

patient variation in pre-course haemoglobin levels and

chemotherapy course number. A summary of haemoglobin

data for individual patients on all courses of chemotherapy,

including those courses on which a research study was

performed, is given in Table I. Additional analysis to

determine the relationship between actual blood volume

taken for research studies and the change in individual patient

Pediatr Blood Cancer DOI 10.1002/pbc

724 Cole et al.

haemoglobin levels showed no significant correlation

between these parameters (Fig. 1). This was the case

regardless of whether the total amount of blood taken was

expressed as an absolute volume of blood or as a percentage

of the total estimated blood volume for an individual child.

The nature of a retrospective study of this type limits

the assessment of actual clinical impact to data, which would

be consistently available in clinical notes. The requirement

for a blood transfusion post-chemotherapy was chosen as a

measure of clinical impact, however, the relationship

between the requirement for blood transfusion and participa-

tion in the pharmacokinetic study proved difficult to assess.

The data were confounded by variation in clinical practice

regarding indications for blood transfusions. The require-

ment for blood transfusion depended, to some extent, on

clinical judgement. In the majority of hospitals involved in

paediatric oncology practice in the UK, blood transfusions

are given when haemoglobin levels fall below 8 g/dl.

However, participation of a child in a research study,

involving the collection of additional blood samples, may

influence the decision to give a transfusion, irrespective of

haemoglobin count. Moreover, the majority of patients were

involved in sampling for research purposes on later

chemotherapy treatment courses. This resulted in differences

in cumulative myelotoxic chemotherapy prior to participa-

tion in the pharmacokinetic study which made it difficult to

determine the impact on the data of cumulative haematolo-

gical toxicity.

These data suggest that participation of patients in a

clinical pharmacology study is associated with a greater fall

in haemoglobin, but that this did not correlate with the total

volume of blood taken for the study. This brings into question

the rationale for the current practice of using an upper limit of

5% of blood volume per 24-hr period for research sampling,

TABLE I. Summary of Haemoglobin Data for all Courses of Chemotherapy Administered toPatients on the Research Study

Patient Age (yrs) Weight (kg)

Course number

1 2 3 4 5 6

1 15 67.9 Pre 13.0 11.2 10.9 10.8 13.2 12.9

Nadir 10.3 8.0 7.2 8.1 7.4 7.9

Diff 2.7 3.2 3.7 2.7 5.8 5.0

2 10 20.9 Pre 12.4 9.9 12.0 13.1 11.7 11.0

Nadir 8.9 7.1 7.0 8.4 5.6 8.0

Diff 3.5 2.8 5.0 4.7 6.1 3.0

3 17 64.0 Pre 10.0 10.5 10.2 9.5 7.9 10.1

Nadir 9.7 9.8 9.5 7.9 NA 8.5

Diff 0.3 0.7 0.7 1.6 NA 1.6

4 7 21.2 Pre 12.1 10.6 9.1 13.8 9.1 12.0

Nadir 10.6 9.1 8.6 9.1 8.4 7.5

Diff 1.5 1.5 0.5 4.7 0.7 4.5

5 13 33.4 Pre 12.4 9.8 8.0 9.6 12.6 9.2Nadir 9.8 8.4 NA 7.9 9.1 7.5Diff 2.6 1.4 NA 1.7 3.5 1.7

6 3 15.1 Pre 10.1 8.3 11.9 9.3 9.8 8.7

Nadir 8.3 8.0 9.3 9.8 8.7 8.6

Diff 1.8 0.3 2.6 �0.5 1.1 0.1

7 19 45.6 Pre NA 12.0 10.0 14.4 14.5 10.0Nadir 12.0 9.6 9.5 7.0 9.9 7.0Diff NA 2.4 0.5 7.4 4.6 3.0

8 1 8.1 Pre 11.4 11.9 15.3 9.8 12.3 9.1

Nadir 10.9 8.5 9.8 7.8 9.1 8.4

Diff 0.5 3.4 5.5 2.0 3.2 0.7

9 1 10.6 Pre 11.3 13.3 10.4 10.2 11.6 NA

Nadir 8.4 10.8 9.0 11.6 8.5 NA

Diff 2.9 2.5 1.4 �1.4 3.1 NA

10 11 30.8 Pre 13.0 10.5 12.2 13.6 9.4 8.8

Nadir 9.9 8.3 8.7 9.4 7.9 NA

Diff 3.1 2.2 3.5 4.2 1.5 NA

11 19 60.0 Pre 12.9 11.4 12.9 12.6 11.2 9.8

Nadir 10.6 9.8 10.8 10.7 9.2 9.6

Diff 2.3 1.6 2.1 1.9 2.0 0.2

Cells in the table contain pre-course levels, cycle nadir and difference between pre-course and nadir levels.

Courses on which a research study was performed are shown in bold. All values represent Hb levels in g/dl.

NA indicates data not available or patient not treated.

Pediatr Blood Cancer DOI 10.1002/pbc

Clinical Impact of Research Studies in Paediatrics 725

and suggests that the factors responsible for a reduction in

haemoglobin need to be investigated further.

Prospective Studies and Ethical Concerns

While this retrospective study has a number of limitations,

the results provide a basis for conducting a well designed

prospective study. Such a study would need to be designed to

control for the many variables that may impact on analysis of

the clinical data. Given the number of confounding variables,

such a studywould require the involvement of a large number

of patients receiving standardised courses of chemotherapy,

with close monitoring of designated clinical markers

following the withdrawal of varying amounts of additional

blood volumes. Whilst there may be ethical challenges in

planning such a study, it is important that concerns related to

the clinical care of children during participation in research

studies are addressed further. Alternatively, routine collec-

tion of haematological data from patients participating in

research studies could be introduced where multiple blood

sampling is involved.

Minimising Sample Number and Volume

With particular relevance to the future conduct of clinical

pharmacology research studies, it is clearly important to

design investigations in such away as to limit the number and

volume of blood samples required. This may involve the

development of drug assays with increased sensitivity, thus

requiring smaller sample volumes for analysis. With recent

developments in analytical methodologies, and wider avail-

ability of techniques such as liquid chromatography—mass

spectroscopy (LC-MS), analysis of an increasing number of

drugs from a small sample volume can be extended in

paediatric oncology. In addition, every effort should be made

to limit the actual number of samples required using

population pharmacokinetics and limited sampling model

approaches [15].

A good example of the use of population pharmaco-

kinetics to optimise drug dosing is provided by the

aminoglycoside antibiotics. Exposure to persistently high

concentrations of aminoglycoside antibiotics such as genta-

micin is associated with toxicity whereas treatment failure

can result from exposure to concentrations below the defined

minimum inhibitory concentration [16]. The clinical rele-

vance of defining and obtaining the optimal therapeutic

exposure is indicated by evidence of the development of

severe septicaemia in febrile neutropenic patients treated

inappropriately with gentamicin [17]. Population pharma-

cokinetic studies involving limited sampling approaches

have been carried out to investigate the efficiency of dosage

nomograms in achieving therapeutic drug concentrations in

patients. This approach has been used to address issues

surrounding variation in gentamicin pharmacokinetics in

patients with cancer [18].

Whilst not common-place in the treatment of children

with chemotherapeutics, population models have been used

successfully in paediatric oncology for drugs such as

melphalan and carboplatin, using limited sampling ap-

proaches based on comprehensive pharmacokinetic data

from larger patient populations. This allows for accurate

estimations of drug exposures, with minimal inconvenience

to the children being treated [19–22].

CONCLUSIONS

When considering any research involving children

with cancer, it is essential that studies are well designed to

answer key scientific questions that may lead to improve-

ments in the clinical use of drugs. The participation of

children with cancer in research studies requires that parents

and often patients themselves give informed consent. This

includes an understanding of any potential adverse outcomes

that may result from participation in the study. It is important

to understand the impact of repeated blood sampling in

children to be able to provide accurate information to parents

and patients and work towards providing clinically relevant

‘guidelines’ for the design of research studies where there is a

requirement for additional blood samples to be taken. It is

anticipated that the development of population pharmacoki-

netics and limited sampling model approaches for an

increasing number of anti-cancer agents used in paediatric

oncology will help to minimise the potential clinical impact

of conducting future clinical pharmacology research studies.

REFERENCES

1. EvansWE, PetrosWP, RellingMV, et al. Clinical pharmacology of

cancer chemotherapy in children. Ped Clin N Am 1989;36:1199–

1230.

0 2 4 6 8 10

Volume of blood (% of total blood volume)

0

2

4

6

Dec

reas

ein

haem

oglo

bin

(g/d

l)

Fig. 1. Relationship between the decrease in haemoglobin count of

patients on a research study, from the pre-course haemoglobin level to

the nadir value, and the volume of blood taken for research purposes

over 3 study days.

Pediatr Blood Cancer DOI 10.1002/pbc

726 Cole et al.

2. Nahata MC. Variability in clinical pharmacology of drugs in

children. J Clin Pharm Therap 1992;17:365–368.

3. Kearns GL, Reed MD. Clinical pharmacokinetics in infants and

children. Clin Pharmacokinetics 1989;17:29–67.

4. McLeod HL, Relling MV, Crom WR, et al. Disposition of

antineoplastic agents in the very young child. Br J Cancer 1992;

66:S23–S29.

5. Woods WG, O’Leary M, Nesbit ME. Life-threatening neuropathy

and hepatotoxicity in infants during induction therapy for acute

lymphoblastic leukemia. J Pediatr 1981;98:642–645.

6. Marsoni S, Ungerleider R, Hurson S, et al. Tolerance to

antineoplastic agents in children and adults. Cancer Treat Rep

1985;69:1263–1269.

7. Crom WR, Relling MV, Christensen ML, et al. Age-related dif-

ferences in hepatic drug clearance in children: Studies with

lorazepam and antipyrine. Clin Pharm Ther 1991;50:132–140.

8. Conroy S, McIntyre J, Choonara I, et al. Drug trials in children:

Problems and the way forward. Br J Clin Pharmacol 2000;49:93–

97.

9. Evans W, Crom W, Abromowitch M, et al. Clinical pharmacody-

namics of high-dose methotrexate in acute lymphocytic leukemia.

Identification of a relation between concentration and effect. N Eng

J Med 1986;314:471–477.

10. EvansWE, RellingMV, Rodman JH, et al. Conventional compared

with individualised chemotherapy for childhood acute lympho-

blastic leukemia. N Eng J Med 1998;338:499–505.

11. Thomas HD, Boddy AV, English MW, et al. Prospective validation

of renal function-based carboplatin dosing in children with cancer:

A United Kingdom Children’s Cancer Study Group trial. J Clin

Oncol 2000;18:3614–3621.

12. Veal GJ, Boddy AV, Thomas HD, et al. Real-time monitoring of

carboplatin pharmacokinetics in paediatric patients receiving high

dose chemotherapy. Br J Cancer 1999;80:93.

13. Kauffman RE. Clinical trials in children. Problems and pitfalls.

Paediatr Drugs 2000;2:411–418.

14. Veal GJ, Dias C, Price E, et al. Influence of cellular factors and

pharmacokinetics on the formation of platinum DNA adducts in

leucocytes of children receiving cisplatin therapy. Clin Cancer Res

2001;7:2205–2212.

15. Panetta JC, Iacono LC, Adamson PC, et al. The importance of

pharmacokinetic limited sampling models for childhood cancer

drug development. Clin Cancer Res 2003;9:5068–5077.

16. Thomson A. Why do therapeutic drug monitoring. Pharmaceutical

J 2004;273:153–155

17. International Antimicrobial Therapy Cooperative Group of the

European Organization for Research and Treatment of Cancer.

Efficacy and toxicity of single daily doses of amikacin and ceftri-

axone versus multiple daily doses of amikacin and ceftazidime for

infection in patients with cancer and granulocytopenia. Ann Intern

Med 1993;119:584–593.

18. Rosario MC, Thomson AH, Jodrell DI, et al. Population

pharmacokinetics of gentamicin in patients with cancer. Br J Clin

Pharmacol 1998;46:229–236.

19. Peng B, Boddy AV, Cole M, et al. Comparison of methods for

the estimation of carboplatin pharmacokinetics in children. Eur J

Cancer 1995;31:1804–1810.

20. Chatelut E, Boddy AV, Peng B, et al. Population pharmacokinetics

of carboplatin in children. Clin Pharmacol Ther 1996;59:436–443.

21. Veal GJ, Foot A, McDowell H, et al. Optimization of high dose

carboplatin treatment in children with Stage IV soft tissue sarcoma

(SIOP/UKCCSG MMT 98 study) by real-time pharmacokinetic

monitoring. Proc Am Assoc Cancer Res 2002;43:1364.

22. Vassal G, Tranchand B, Valteau-Couanet D, et al. Pharmacody-

namics of tandem high-dose melphalan with peripheral blood stem

cell transplantation in children with neuroblastoma and medullo-

blastoma. Bone Marrow Transplant 2001;27:471–477.

Pediatr Blood Cancer DOI 10.1002/pbc

Clinical Impact of Research Studies in Paediatrics 727