- Original research
- Open Access
Pharmacokinetic modeling of [18F]fluorodeoxyglucose (FDG) for premature infants, and newborns through 5-year-olds
© Khamwan et al. 2016
- Received: 20 December 2015
- Accepted: 29 February 2016
- Published: 17 March 2016
Absorbed dose estimates for pediatric patients require pharmacokinetics that are, to the extent possible, age-specific. Such age-specific pharmacokinetic data are lacking for many of the diagnostic agents typically used in pediatric imaging. We have developed a pharmacokinetic model of [18F]fluorodeoxyglucose (FDG) applicable to premature infants and to 0- (newborns) to 5-year-old patients, which may be used to generate model-derived time-integrated activity coefficients and absorbed dose calculations for these patients.
The FDG compartmental model developed by Hays and Segall for adults was fitted to published data from infants and also to a retrospective data set collected at the Boston Children’s Hospital (BCH). The BCH data set was also used to examine the relationship between uptake of FDG in different organs and patient weight or age.
Substantial changes in the structure of the FDG model were required to fit the pediatric data. Fitted rate constants and fractional blood volumes were reduced relative to the adult values.
The pharmacokinetic models developed differ substantially from adult pharmacokinetic (PK) models which can have considerable impact on the dosimetric models for pediatric patients. This approach may be used as a model for estimating dosimetry in children from other radiopharmaceuticals.
- Pediatric imaging
- Compartmental modeling
The radiation exposure resulting from medical imaging has become a public safety concern [1–3]. Dose reduction for pediatric patients is particularly important since such patients are considered to be at increased risk for cancer owing to the enhanced radiosensitivity of their tissues and the longer time period over which stochastic radiation effects may manifest [4, 5].
Guidelines on the amount of activity to administer for pediatric nuclear medicine imaging are based on expert consensus of best practices [6, 7]. Methods based on balancing activity administration with whole-body photon fluence or diagnostic image quality to arrive at an optimal administered activity have also been examined [8–11]. Optimization efforts would benefit by the availability of pharmacokinetic data for radiopharmaceuticals commonly used in pediatric nuclear medicine imaging. An extensive set of absorbed dose estimates and corresponding pharmacokinetic data has been published by the International Commission on Radiological Protection (ICRP) for many radiopharmaceuticals [12, 13]. The tabulated calculations include absorbed and effective doses to children. The biokinetic models used in these calculations, however, are typically derived from adult data, and the applicability of these models to children has not been ascertained. There are a number of studies that provide pharmacokinetic (PK) data for fluorodeoxyglucose (FDG) in pediatric patients [14–19]. Few to none of these studies, however, include PK data for tissues other than brain and, in one case, bladder . In this work, we derive an [18F]-FDG model for early-age pediatric patients (newborns to 5-year-olds) based on an established [18F]-FDG model applicable to adults , which is made applicable to pediatric patients by adjusting the model and fitting it to a combination of published data and retrospective data collected at Boston Children’s Hospital (BCH). The latter data set was also used to examine the relationship between FDG uptake in different organs and patient weight or age. Such relationships will be useful as input into image simulation and diagnostic image quality evaluation tasks as described previously .
To arrive at a pharmacokinetic FDG model applicable to pediatric patients, we started with a published FDG PK model applicable to adults . Using data from the literature  and a data set from BCH, the model was adjusted and used to fit the combined measured and literature-derived data set. In consultation with the institutional review board (IRB), the use of already collected, anonymized, imaging data for the purposes of this study was deemed exempt from IRB review.
Newborns to 5-year-olds FDG pharmacokinetic data
Patients’ characteristic data
Acquisition time after injection (min)
Provisional/suspected diagnosis (reason for PET/CT examination)
Neurofibromatosis 1 (abdominal/right flank pain)
Suspected pelvic carcinoma
Retroperitoneal sarcoma s/p chemotherapy (assess for tumor activity)
Suspected left scalp Ewing’s sarcoma
Suspected malignant liver lesionb
Rhabdomyosarcoma s/p therapya
Rhabdomyosarcoma s/p therapyb
Rhabdomyosarcoma s/p therapyb
Infantile fibrosarcoma of pelvis s/p therapya
Diffuse large B cell lymphoma stage IVa
Hodgkin’s lymphoma s/p chemotherapya
Localized Ewing’s sarcomaa
Neurofibromatosis 1 (tumor activity evaluation)
Suspected malignant pelvic massb
High-risk neuroblastoma s/p therapya
Multifocal inflammatory myofibroblastic tumor of the lungs s/p chemotherapya
B cell lymphomaa
Anaplastic large-cell lymphoma s/p therapya
Hepatoblastoma s/p chemotherapya
Suspected left renal cell carcinomaa
Right calf alveolar rhabdomyosarcomaa
Stage IV neuroblastoma s/p therapya
Stage IV neuroblastoma s/p therapya
Suspected neck LN in thyroid cancerb
Metastatic glomus tumora
Malignant rhabdoid tumor s/p therapya
FDG compartmental model for premature infants
A compartmental modeling package, SAAM II (The Epsilon Group, Charlottesville, VA), was used for model fitting . We used the whole-body adult FDG pharmacokinetic compartment model developed by Hays and Segall  and fitted it to partial data collected from infants. The premature infant pharmacokinetic data were derived from a report published by Niven and Nahmias . In brief, these authors collected two consecutive 45-min dynamic PET scans in very low birth weight infants. The first scan was over the head, and the second was over the chest region. The time-activity curves for the brain, heart wall, lungs, and kidneys were then generated.
Model fitting to the premature infant data was obtained by adjusting the adult model parameters of each compartment that directly exchanges FDG with the plasma. The exchange rate between the plasma and erythrocyte compartments was also adjusted in this initial fitting phase. These initial fits were performed using brain FDG exchange values obtained from Huang et al.  which consisted of gray matter and white matter with bidirectional exchange of FDG between plasma and rapidly and slowly exchanging FDG compartments. We eliminated the distinction between white and gray matter and only retained the distinction between rapidly and slowly exchanging brain compartments. The fraction of blood volume was also gradually adjusted in order to obtain the best fit of the brain compartment model.
In fitting the lungs and heart wall, we expanded the model from a single compartment sink to two compartments that exhibit bidirectional exchange of FDG with the plasma. The fraction of blood volume in each compartment was also adjusted. The urine compartment in the adult FDG model was modified to represent the kidneys, and a bidirectional exchange with the plasma was added. The bidirectional rate constants between the plasma and kidneys were also adjusted. The compartmental structure associated with the liver and other tissues was retained as described in the adult FDG model.
FDG compartmental model for 0- to 5-year-olds
To create the pediatric (newborns to 5-year-olds) model, the FDG model of Hays and Segall was initially fitted to the pharmacokinetic data reported by Niven and Nahmias, as described above, and then to the data obtained from BCH. The compartmental structures were kept in accordance with the infant model. As the acquisition time spanned a range between 60 and 126 min after injection for 35 patients, the data were binned into 5-min intervals and the mean and a standard deviation for the data falling into each bin was calculated and used as part of the model fitting process. Human FDG biodistribution data at multiple time points are not available for pediatric patients. As a result, the data obtained from multiple patients spanning different acquisition times were fitted into the model. We adjusted the transfer rate constant parameters between the compartment gradually for the brain, lungs, heart wall, kidneys, and liver to fit the model to these data. The SAAM II software will then generate time-integrated activity curve of the model fitted to the observed data based on a nonlinear least-squares regression algorithm. The blood volume fraction in each compartment, representing the blood physically contained in an organ or tissue relative to the total-body blood volume, was also changed. The kinetic parameters associated with rapidly and slowly exchanging tissue compartments were retained as in the infant model; however, the bidirectional exchange of the FDG between the plasma and erythrocytes had to be adjusted. The differential equations and parameter definitions describing both models are provided in Additional file 1.
Fits to organ concentration vs weight
where a, b, c, d are the fitted parameter values. Binning the data in two different time-interval lengths would be useful for observing the different results of the percent injected activity of the FDG uptake in each organ at the early time (60–81 min) and later time (82–126) period for generating the image simulation in the future study.
Parameter values fitted to the premature infants and newborns to 5-year-olds FDG model
0- to 5-year-olds
Plasma to erythrocytes (k1)
Erythrocytes to plasma (k2)
Plasma to fast brain (k3)
Fast brain to plasma (k4)
Fast brain to slow brain (k5)
Slow brain to fast brain (k6)
Plasma to lungs (k7)
Lungs to plasma (k8)
Plasma to heart wall (k9)
Heart wall to plasma (k10)
Plasma to kidneys (k11)
Kidneys to plasma (k12)
Plasma to fast liver (k13)
Fast liver to plasma (k14)
Fast liver to slow liver (k15)
Plasma to fast “other” (k16)
Fast “other” to plasma (k17)
Fast “other” to slow “other” (k18)
Blood volume fraction in brain
Blood volume fraction in lungs
Blood volume fraction in heart
Blood volume fraction in liver
Time-integrated activity coefficient (TIAC) derived from the newborn FDG model compared with the published data
Hays and Segall
Niven and Nahmias
0- to 5-year-olds
2.20E−01 ± 0.09
2.82E−01 ± 0.07
7.00E−02 ± 0.03
4.80E−02 ± 0.03
1.30E−01 ± 0.06
1.80E−02 ± 0.01
1.20E−02 ± 0.01
1.50E−01 ± 0.05
List of percent blood volume predicted by the FDG newborn model compared with values for the adult in ICRP 106 and the original FDG adult model
Fraction blood volume (%)
Adults (ICRP 106)
Adults (Hays and Segall)
0- to 5-year-olds
1.0 (same listed has coronary tissue)
6.9 (includes coronary artery)
Fitting coefficients used to determine the %IA/g for each organ at different time point for newborn FDG model
The goal for every pediatric molecular imaging study is to obtain the best diagnostic information employing the highest quality standards, in the shortest period of time, and with the lowest patient radiation exposure . The Image Gently Campaign, an initiative of the Alliance for Radiation Safety in Pediatric Imaging, has highlighted the need to tailor diagnostic imaging procedures to children so as to reduce their radiation exposure and potential cancer risk (http://www.imagegently.org, accessed May 2015).
Almost all of the pharmacokinetic measurements available for absorbed dose and risk calculations are based on data collected from adults . Using data from the literature and from retrospective measurements in different patients obtained from BCH, we have developed pharmacokinetic models for dosimetry and activity concentration as a function of body weight to be used for image simulation. Due to incomplete descriptions of acquisition parameters and possible differences in sensitivity, it is difficult to compare the data obtained from the literature with the retrospective data we collected at BCH . Accordingly, we have superimposed the data from the literature, when available, with BCH data and model fits in Fig. 3 to highlight the serious need for a consistent PK data set for pediatric patients. Also, the BCH data are at a single nominal point in time but, due to the practicalities associated with imaging pediatric patients, there was a substantial variability in imaging time. This allowed us to generate kinetic data over the relative short time span defined by the BCH data set. A more comprehensive data collection effort would require an imaging protocol to image at additional time points. Finally, as shown in Table 1, each time point is derived from a single patient. Given the limitations associated with pediatric imaging, it may be difficult for a series of pediatric patients to be imaged over multiple time points; rather, data from multiple patients spanning different acquisition times will need to be assembled to establish a pharmacokinetic profile for FDG and other agents used in pediatric imaging.
A number of interesting observations may be extracted from the results presented above. We find that the %IA in the brain obtained from BCH data is greater than predicted from the adult model and also from the premature infant data. Correspondingly, the brain TIAC derived from the BCH data is approximately four times greater than the other estimates shown in Table 3. The TIAC in heart wall was about fivefold lower in the premature infants than the value for adults calculated using the Hays and Segall FDG model. Brain and lung AUC values for the fitted premature infant model were about the same as those in the adult FDG model. The TIACs obtained from fits to the BCH data (0- to 5-year-olds) differ from the premature infant fits as might be expected given the nature of the premature infant data in which these patients were being imaged due to lung infections. Accordingly, the lung TIAC in this patient population is 3.7-fold higher than that seen in the BCH data. The brain, heart wall, and kidneys are 76, 16, and 76 % higher, respectively, in the BCH data set compared to the Niven and Nahmias data set.
In Table 4, the model-derived estimates of the percent blood volume for the brain, heart wall, lungs, and liver are compared with published values. The fitted values for both the premature infants and for the BCH data set are greater for brain and heart wall relative to the values for adults reported in ICRP 106. The fractional blood volume for liver derived from the BCH data sets is the same as that reported for adults. Liver data were not reported by Niven and Nahmias. The fractional blood volume in lungs of premature infants and newborns to 5-year-old children are 40 % and fourfold lower, respectively, than the adult values reported in ICRP 106. The model-derived premature infant heart wall percent tissue blood volume is greater than in the ICRP 106 reference adult by more than a factor of two. The Hays and Segall value for heart wall is about three times greater than the value we obtained for the newborns. The newborn lung percent blood volume is lower than the value listed in ICRP 106 and in the adult FDG model. These comparisons, especially for the heart, are made difficult because of the equivocal descriptions of blood content and region described. For example, Niven and Nahmias described a region of interest over the heart for imaging-based measurements, which would presumably include both parenchymal (heart wall blood) and heart contents. The calculated TIAC, however, is ascribed to heart wall based on the assumption that little activity would be in the blood after 45 min, the time of imaging. The Hays and Segall paper provides a footnote to the listed value of percent blood volume indicating that the fractional blood volume includes blood in the coronary artery content.
Current pediatric absorbed dose estimates are performed using adult pharmacokinetic data with S values that account for the anatomical differences between adults and children. The divergence, in both methodology, patient population and results obtained amongst the different available sources of data for pediatric pharmacokinetic modeling of FDG, highlights the need for greater data collection of pediatric imaging agents.
Model-derived extrapolation of adult pharmacokinetic data provides an initial approach to extending pediatric PK data for use in dosimetry and image simulation. Additional measurements over time are needed to further validate these pediatric FDG models.
This work was supported by NIH grant R01 EB013558.
Compliance with ethical standards
This study was funded by NIH grant R01 EB013558.
This article does not contain any studies with human participants performed by any of the authors.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.
- Schauer DA, Linton OW. NCRP report No. 160, ionizing radiation exposure of the population of the United States, medical exposure—are we doing less with more, and is there a role for health physicists? Health Phys. 2009;97(1):1–5.View ArticlePubMedGoogle Scholar
- Hricak H, Brenner DJ, Adelstein SJ, Frush DP, Hall EJ, Howell RW, et al. Managing radiation use in medical imaging: a multifaceted challenge. Radiology. 2011;258(3):889–905.View ArticlePubMedGoogle Scholar
- Mettler FAJ, Thomadsen BR, Bhargavan M, Gilley DB, Gray JE, Lipoti JA, et al. Medical radiation exposure in the US in 2006: preliminary results. Health Phys. 2008;95(5):502–7.View ArticlePubMedGoogle Scholar
- National Research Council. Health risks from exposure to low levels of ionizing radiation: BEIR VII—phase 2. Washington, DC: National Research Council; 2005.Google Scholar
- ICRP. Publication 103: the 2007 recommendations of the International Commission on Radiological Protection. Ann ICRP. 2007;37(2-4):1–332.View ArticleGoogle Scholar
- Lassmann M, Treves ST. Paediatric radiopharmaceutical administration: harmonization of the 2007 EANM paediatric dosage card (version 1.5. 2008) and the 2010 North American consensus guidelines. Eur J Nucl Med Mol Imaging. 2014;41(5):1036–41.View ArticlePubMedGoogle Scholar
- Treves ST, Lassmann M. International guidelines for pediatric radiopharmaceutical administered activities. J Nucl Med. 2014;55(6):869–70.View ArticlePubMedGoogle Scholar
- Lassmann M, Biassoni L, Monsieurs M, Franzius C. The new EANM paediatric dosage card: additional notes with respect to F-18. Eur J Nucl Med Mol Imaging. 2008;35(9):1666–8.View ArticlePubMedGoogle Scholar
- Lassmann M, Biassoni L, Monsieurs M, Franzius C, Jacobs F. The new EANM paediatric dosage card. Eur J Nucl Med Mol Imaging. 2007;34(5):796–8.View ArticlePubMedGoogle Scholar
- Sgouros G, Frey EC, Bolch WE, Wayson MB, Abadia AF, Treves ST. An approach for balancing diagnostic image quality with cancer risk: application to pediatric diagnostic imaging of 99mTc-dimercaptosuccinic acid. J Nucl Med. 2011;52(12):1923–9.View ArticlePubMedPubMed CentralGoogle Scholar
- Alessio AM, Sammer M, Phillips GS, Manchanda V, Mohr BC, Parisi MT. Evaluation of optimal acquisition duration or injected activity for pediatric 18F-FDG PET/CT. J Nucl Med. 2011;52(7):1028–34.View ArticlePubMedGoogle Scholar
- ICRP. Publication 53: radiation dose to patients from radiopharmaceuticals. Ann ICRP. 1988;18(1-4):1–377.Google Scholar
- ICRP. Publication 106: radiation dose to patients from radiopharmaceuticals: addendum 3 to ICRP publication 53. Ann ICRP. 2008;38(1-2):1–197.View ArticlePubMedGoogle Scholar
- Hua C, Merchant TE, Li X, Li Y, Shulkin BL. Establishing age-associated normative ranges of the cerebral 18F-FDG uptake ratio in children. J Nucl Med. 2015;56(4):575–9.View ArticlePubMedGoogle Scholar
- Ruotsalainen U, Suhonen-Polvi H, Eronen E, Kinnala A, Bergman J, Haaparanta M, et al. Estimated radiation dose to the newborn in FDG-PET studies. J Nucl Med. 1996;37(2):387–93.PubMedGoogle Scholar
- London K, Howman-Giles R. Normal cerebral FDG uptake during childhood. Eur J Nucl Med Mol Imaging. 2014;41(4):723–35.View ArticlePubMedGoogle Scholar
- Chugani HT, Phelps ME. Maturational changes in cerebral function in infants determined by 18F-FDG positron emission tomography. Science. 1986;231(4740):840–3.View ArticlePubMedGoogle Scholar
- Chugani HT, Phelps ME, Mazziotta JC. Positron emission tomography study of human brain functional development. Ann Neurol. 1987;22(4):487–97.View ArticlePubMedGoogle Scholar
- Van Bogaert P, Wikler D, Damhaut P, Szliwowski H, Goldman S. Regional changes in glucose metabolism during brain development from the age of 6 years. Neuroimage. 1998;8(1):62–8.View ArticlePubMedGoogle Scholar
- Hays MT, Segall GM. A mathematical model for the distribution of fluorodeoxyglucose in humans. J Nucl Med. 1999;40(8):1358–66.PubMedGoogle Scholar
- Niven E, Nahmias C. Absorbed dose to very low birth weight infants from 18F-fluorodeoxyglucose. Health Phys. 2003;84(3):307–16.View ArticlePubMedGoogle Scholar
- Geyer AM, O’Reilly S, Lee C, Long DJ, Bolch WE. The UF/NCI family of hybrid computational phantoms representing the current US population of male and female children, adolescents, and adults—application to CT dosimetry. Phys Med Biol. 2014;59(18):5225–42.View ArticlePubMedGoogle Scholar
- Barrett PHR, Bell BM, Cobelli C, Golde H, Schumitzky A, Vicini P, et al. SAAM II: simulation, analysis, and modeling software for tracer and pharmacokinetic studies. Metabolism. 1998;47(4):484–92.View ArticlePubMedGoogle Scholar
- Huang SC, Phelps ME, Hoffman EJ, Sideris K, Selin CJ, Kuhl DE. Noninvasive determination of local cerebral metabolic rate of glucose in man. Am J Physiol. 1980;238(1):E69–82.PubMedGoogle Scholar
- Hänscheid H, Fernández M, Lassmann M. The absorbed dose to blood from blood-borne activity. Phys Med Biol. 2015;60(2):741–53.View ArticlePubMedGoogle Scholar
- Treves ST. Pediatric nuclear medicine/PET. 3rd ed. New York NY: Springer Science + Business Media LLC; 2007.View ArticleGoogle Scholar
- Boellaard R. Standards for PET image acquisition and quantitative data analysis. J Nucl Med. 2009;50 Suppl 1:11S–20.View ArticlePubMedGoogle Scholar