Original Article

Cost Effectiveness of Proton Therapy Compared With Photon

Therapy in the Management of Pediatric Medulloblastoma
Raymond B. Mailhot Vega, MD, MPH1; Jane Kim, PhD2; Marc Bussière, MS3; Jona Hattangadi, MD4; Abby Hollander, MD5;
Jeff Michalski, MD1; Nancy J. Tarbell, MD3; Torunn Yock, MD, MPH3; and Shannon M. MacDonald, MD3

BACKGROUND: Proton therapy has been a hotly contested issue in both scientific publications and lay media. Proponents cite the
modality’s ability to spare healthy tissue, but critics claim the benefit gained from its use does not validate its cost compared with
photon therapy. The objective of this study was to evaluate the cost effectiveness of proton therapy versus photon therapy in the
management of pediatric medulloblastoma. METHODS: A cost-effective analysis was performed from the societal perspective using a
Monte Carlo simulation model. A population of pediatric medulloblastoma survivors aged 18 years was studied who had received
treatment at age 5 years and who were at risk of developing 10 adverse events, such as growth hormone deficiency, coronary artery
disease, ototoxicity, secondary malignant neoplasm, and death. Costing data included the cost of investment and the costs of diagno-
sis and management of adverse health states from institutional and Medicare data. Longitudinal outcomes data and recent modeling
studies informed risk parameters for the model. Incremental cost-effectiveness ratios were used to measure outcomes. RESULTS:
Results from the base case demonstrated that proton therapy was associated with higher quality-adjusted life years and lower costs;
therefore, it dominated photon therapy. In 1-way sensitivity analyses, proton therapy remained the more attractive strategy, either
dominating photon therapy or having a very low cost per quality-adjust life year gained. Probabilistic sensitivity analysis illustrated
the domination of proton therapy over photon therapy in 96.4% of simulations. CONCLUSIONS: By using current risk estimates and
data on required capital investments, the current study indicated that proton therapy is a cost-effective strategy for the management
of pediatric patients with medulloblastoma compared with standard of care photon therapy. Cancer 2013;119:4299-307.
C 2013 American Cancer Society.
V

KEYWORDS: medulloblastoma, cost effectiveness, comparative effectiveness, proton, Monte Carlo.

INTRODUCTION
Proton therapy is considered one of the most advanced modalities for radiation therapy (RT), with properties that allow
for the avoidance of radiation to healthy, uninvolved tissues outside of the region that requiring radiation for tumor con-
trol.1 By contrast, photon RT, which is more widely available, can deliver an unnecessary dose to healthy tissues, carrying
with it a risk of various adverse events (Fig. 1). Reports of long-term clinical outcomes of proton RT are not yet available
for most malignancies,2 and the construction of a proton center requires a capital investment of approximately $140 mil-
lion, with a cost to society of treatment for an average proton radiation course estimated at $40,000.3,4 Many recently
published articles and editorials have addressed the lack of evidence for the cost effectiveness of proton therapy with regard
to specific malignancies,5 particularly prostate cancer. In the current health care climate, it is necessary that we closely ana-
lyze whether proton radiation truly delivers superior clinical outcomes and cost benefits compared with standard photon
radiation. Despite significantly higher capital and operational costs, proton therapy may prove to be cost effective for chil-
dren,6 who are particularly susceptible to radiation-induced side effects.
Medulloblastoma is the most common malignant cancer of the central nervous system in children, and RT is consid-
ered the standard of care for any child aged >3 years.7 Regimens that incorporate radiation result in an overall 5-year sur-
vival rate of 75% to 85%.8 These rates have improved substantially over the past decade.9 Thus, as cancer survival rates
have improved, the impact of treatment-related side effects has become even more important. Chronic medical problems
induced by RT can impact the quality of life of these patients and are costly to families, insurers, and society. Survivors of
childhood medulloblastoma are at risk of multiple adverse events because of radiation: cognitive deficits, growth hormone

Corresponding author: Shannon M. MacDonald, MD, Harvard Medical School, Massachusetts General Hospital, 100 Blossom Street, Boston, MA 02114; Fax: (617)
726.3603; smacdonald@partners.org.
1
Department of Radiation Oncology, Washington University in St. Louis School of Medicine, St. Louis, Missouri; 2Harvard School of Public Health, Boston,
Massachusetts; 3Department of Radiation Oncology, Harvard Medical School, Massachusetts General Hospital, Boston, Massachusetts; 4Department of Radiation
Oncology, University California San Diego, San Diego, California; 5Department of Pediatric Endocrinology, Washington University in St. Louis School of Medicine,
St. Louis, Missouri

We would like to acknowledge Anthony Oliveira for assistance in obtaining cost data.

DOI: 10.1002/cncr.28322, Received: May 11, 2013; Revised: July 7, 2013; Accepted: July 11, 2013, Published online September 16, 2013 in Wiley Online Library
(wileyonlinelibrary.com)

Cancer December 15, 2013 4299

Original Article

Figure 1. These images compare (Left) proton therapy and (Right) photon therapy plans for cerebrospinal irradiation. The dose
sparing anterior to the spinal column in the proton plan is evident compared with the photon plan. (Middle) The dose of radiation
is delineated in centigrays according to color shade. Courtesy of Judith Adams.

deficiency (GHD), ototoxicity, hypothyroidism, conges-
tive heart failure (CHF), pulmonary disease, coronary ar-
tery disease (CAD), adrenocorticotropic hormone
(ACTH) deficiency, gonadotropin deficiency, early pu-
berty, premature ovarian failure, and radiation-induced
cancers.10 Reports have indicated that adverse events may
affect >55% of survivors.11
The lengthy side-effect profile associated with cra-
niospinal photon RT renders pediatric medulloblastoma
an attractive malignancy for the use of proton therapy.
The use of proton therapy may prevent lifelong chronic
diseases through the avoidance of irradiating a large vol-
ume of normal tissue that otherwise would receive radia-
tion and by sparing a substantial amount of brain from
receiving higher doses of radiation. In addition, the high
necessary capital investment in proton therapy may be
reasonable considering the prolonged clinical benefits in
children.3 This amortization may make proton therapy
more cost effective than conventional photon therapy. To
date, only 1 cost-effective analysis (CEA) of pediatric
medulloblastoma has been published—a Swedish model-
based analysis that identified proton therapy as the more
cost-effective strategy compared with photon therapy,12
which was considered the standard of care. Contemporary
insights into RT and risk models have provided more pre-
cise risk predictions,7 and recently released phase 2 trial
outcomes have provided primary data to enable more
informed decisions.13 With this advancement in knowl-
edge regarding the risks and benefits associated with pro-
ton therapy, the objective of the current study was to
evaluate the cost effectiveness of proton therapy compared
with photon therapy in treating pediatric medulloblas-
toma, incorporating the best available data and costs.
MATERIALS AND METHODS
A first-order Monte Carlo simulation model of survivors
of pediatric medulloblastoma was developed to evaluate
the cost effectiveness of proton therapy versus photon
therapy. Although we assumed that patients received
treatment at age 5 years, we elected to track the health
benefits and costs of individual patients starting at age 18
years to capture the health benefits throughout the adult
life of surviving patients. This modeling strategy allowed
for the appropriate assignment of utilities and avoided

4300 Cancer December 15, 2013

 CEA of Proton vs Photon RT for Medullo/Mailhot Vega et al

TABLE 1. Management Strategies and Sources

State Management Strategy Source

GHD Every 6 mo: endocrinologist visit, serum IGF-1, and fasting glucose; annually: lipid Baskin 200924
profile; routine daily: growth hormone injection
Hypothyroidism Annual TSH and free T4 test; routine daily, levothyroxine Baskin 200225
Ototoxicity Hearing aid with 4-y lifespan, audiometric evaluation, and telephone amplifier with Mohr 200026
10-y lifespan
ACTH deficiency Routine daily, hydrocortisone Grossman 201027
Gonadotropin deficiency Men, daily testosterone and 6-mo serum testosterone checks; women, estradiol Snyder 2012,28
and medroxyprogesterone until age 50 y Nelson & Calis 201229
Heart failure At diagnosis: CBC, UA, serum electrolytes with Ca and Mg, BUN, Cr, fasting blood Hunt 200930
glucose, lipid profile, LFTs, TSH, 12-lead EKG, CXR, and 2D echo with Doppler;
routine daily: ramipril and chlorthalidone
CAD At diagnosis: rest EKG, stress EKG, Hgb, fasting glucose, and fasting lipid panel Fraker 200741
with total cholesterol, HDL, TGs, and LDL; annuals stress EKG; routine daily:
ramipril, simvastatin, metoprolol, and ASA

Abbreviations: 2D, 2-dimensional; ACTH, adrenocorticotropic hormone; ASA, aspirin; BUN, blood urea nitrogen; CAD, coronary artery disease; CBC, complete
blood count; Cr, creatinine; CXR, chest x-ray; EKG, electrocardiogram; GHD, growth hormone deficiency; HDL, high-density lipoprotein; Hgb, hemoglobin;
IGF-1, insulin-like growth factor 1; LDL, low-density lipoprotein; LFTs, liver function tests; TGs, triglycerides; TSH, thyroid-stimulating hormone; UA, urinalysis.

their inappropriate use for pediatric-aged patients, for
whom data are scarce. Each year, individuals in the model
were at risk of entering the following health states: GHD,
ototoxicity, hypothyroidism, CHF, CAD, ACTH defi-
ciency, gonadotropin deficiency, secondary malignant
neoplasm (SMN), relapse, and death. Cognitive deficit
was not included, because no data exist to appropriately
assign a cost to an intelligence quotient (IQ) loss second-
ary to cranial RT. Patients had the potential to experience
different health states each cycle until they reached age
100 years or experienced death. One cycle was equivalent
to 1 year of time.
Each health state carried its own risk, associated util-
ity, and cost; death could occur at any cycle because of
background mortality or excess disease-specific mortality.
Background annual mortality risk in the base case was
derived from standard 2007 life tables.14 Other adverse-
event parameter values were identified through an extensive
literature search. Proton and photon risks for ototoxicity,
hypothyroidism, CHF, CAD, and SMN were derived
from modeling estimates by Brodin et al.7 Reports by Yock
and colleagues13 and Merchant et al15 provided data to esti-
mate the risk of GHD; the publications by Frange et al and
Heikens et al informed the photon risks of ACTH defi-
ciency and gonadotropin deficiency, respectively16,17; and
proton equivalents for those states were derived in concord-
ance with the reports by Miralbell et al and Lundkvist
et al.12,18 Health state utilities were used to reflect dimin-
ished quality of life because of disease or events and were
measured on a scale from 0 (worst health state, usually
death) to 1 (perfect health). The report by Sullivan and
Ghushchyan provided utility values,19 except for ACTH
deficiency,20 GHD,21 and gonadotropin deficiency.22
Costs in the model included the cost of RT and the
cost of managing adverse events. The cost of RT reflected
the capital investment and operational management costs
for working facilities and included estimates for building
and infrastructure, hardware, dosimetry and engineering
equipment, planning and clinical management software,
and necessary licenses as well as working capital during
the construction and transition phases. Salaries and bene-
fits for physicians and staff and hospital overhead also
were captured in the RT cost. These parameters were used
to generate a cost per hour per room, assuming a 10-hour
treatment day and a 40-year facility lifespan, excluding
holidays and weekends. According to Packer et al, 31 frac-
tions were expected, including 13 fractions of 60-minute
cerebrospinal irradiation and 18 fractions of 20-minute
posterior fossa boost.23 The cost per room per hour multi-
plied by 19 hours of room use represented the cost for the
operational and capital costs to manage 1 patient with pe-
diatric medulloblastoma in 2012. For each incorporated
health state, management strategies were determined for
cost estimation from published guidelines, and these are
listed in Table 1. Procedure allowables were valued using
Current Procedural Terminology codes obtained from
the Centers for Medicare and Medicaid Services, and spe-
cific drug management and regimen costs were calculated
using the Red Book and were converted to 2012 values by
using an inflation calculator from the US Department of
Labor.32,33 The model valued health effects and costs
from the societal perspective, as recommended by the US
Panel on Cost Effectiveness in Health and Medicine.34
From the same recommendation, all costs and effects were
discounted at 3% per year assuming treatment at age 5
years. A diagram of the model is illustrated in Figure 2.

Cancer December 15, 2013 4301

Original Article

Figure 2. This diagram illustrates the possible health states (circles) a healthy survivor confronts in the model. The survivor has
the potential to enter more than 1 discrete clinical state in a year: eg, the survivor has the potential to develop growth hormone
deficiency (GHD) and congestive heart failure (CHF) in year 1. In this sample outcome, a patient experiences GHD during year 1,
continues to persist in a GHD health state, and develops coronary artery disease (CAD) in year 2. Note that death is possible in
all stages. Because of the high number of combinations, not all possibilities could be represented pictorially; the ellipses repre-
sent the other possible combinations. SMN indicates secondary malignant neoplasm.

The model projected 2 outcome measures: lifetime
costs expressed in US dollars (USD) and health benefits
expressed in quality-adjusted life years (QALYs). Because
each health state is associated with a particular cost and
quality-of-life weight (ie, utility), a total cost and sum of
QALYs could be calculated for the entire lifespan of an
individual in the model, depending on time spent in each
state. Owing to differing costs and risks of health states,
the 2 model arms of proton therapy and photon therapy
generated different sums. A quotient obtained from the
difference between strategy costs divided by the difference
in QALYs was used to derive an incremental cost-
effectiveness ratio (ICER), which is equivalent to the
incremental cost per additional QALY. The ICER can be
compared with a society’s generally accepted benchmark
of willingness to pay (WTP) for 1 QALY. This model
assumed a societal WTP of $50,000 per QALY.35
One-way sensitivity analyses were conducted to
demonstrate the influence of parameters. In these analy-
ses, lifetime risks were tested in the model by rerunning
simulations with the proton risk equal to the photon risk.
The most influential parameters identified from 1-way
sensitivity analysis were selected for testing in probabilistic
sensitivity analysis (PSA), in which multiple model
parameters are simultaneously varied with set distribu-
tions and ranges; such analyses help illustrate the robust-
ness of the model in light of the joint uncertainty for
model parameters. Risks selected for PSA that were based
on actual patient data were tested with beta distributions.
Costs and risks derived from models were chosen to have
uniform distributions, because those parameters were
derived from models and, thus, have more uncertainty.
RESULTS
Base-case parameters are displayed in Table 2, and the
results from a base-case analysis with 1,000,000 individual
patients are provided in Table 3. The most common events
for the photon arm were hearing loss and secondary cancer,
and GHD and CAD were the most common for the pro-
ton arm. In the base case, proton therapy was more effective
yet less costly compared with photon therapy: we defined
this as proton therapy dominating photon therapy. The
robustness of the base-case results was demonstrated in 1-
way sensitivity analyses, as indicated in Table 4. In all sce-
narios, proton therapy either dominated photon therapy or
was associated with a very low ICER (<$5000 per QALY).
In the sensitivity analysis, risk of hearing loss, risk of
SMN, and risk of heart failure were most influential on

4302 Cancer December 15, 2013

 CEA of Proton vs Photon RT for Medullo/Mailhot Vega et al

TABLE 2. Base Case Model Parameters

Health States

Variable GHD Gonadotropin Deafness Hypothyroid ACTH CAD CHF Second Cancer

Lifetime riska 0.13 (0.24) 0.02 (0.17) 0.1 (0.62) 0.08 (0.3) 0.02 (0.13) 0.14 (0.24) 0.03 (0.09) 0.07 (0.45)
Utility 0.81 0.8035 0.776 0.821 0.91 0.695 0.636 0.795
Annual cost, $ 13,823 2,050b 1,038 158 314 3,362c 940c 39,130

Abbreviations: ACTH, adrenocorticotropic hormone; CAD, coronary artery disease; CHF, congestive heart failure; GHD, growth hormone deficiency.
a
Values displayed for lifetime risk are for proton therapy (photon therapy).
b
After age 50 years, the model input was $1952.
c
The input costs for the first year of diagnosis were $3419 and $1394 for CAD and CHF, respectively.

TABLE 3. Base Case Results

Variable Photon Therapy Proton Therapy Difference
Total QALYs 13.91 17.37 3.46
Total costs, $ 112,789.87 80,210.79 232,579.08
ICER — — Proton dominates
Abbreviations: ICER, incremental cost-effectiveness ratio; QALY, quality-
adjusted life year.
the incremental effectiveness of proton therapy. The cost
of capital investment and the risk of GHD were most in-
fluential on the incremental cost of therapy. These param-
eters were selected for further testing in a PSA for 25,000
patients in 1000 separate samples. A scatter plot depicting
incremental cost versus incremental effectiveness for each
of the 1000 runs is illustrated in Figure 3. All simulations
demonstrated ICERs comparing proton and photon ther-
apy below the WTP threshold of $50,000 per QALY. In
96.4% of the simulations, proton therapy dominated
photon therapy and was cost effective in 100% of
simulations.
DISCUSSION
This study contributes to the limited assessment of the
cost effectiveness of proton therapy in the management of
pediatric medulloblastoma. The results of the base case
demonstrated that proton RT represents good value for
money compared with photon RT. Despite the increased
upfront cost of proton therapy, we observed that protons
were cost saving compared with photons over the lifespan
of a population of patients because of decreased costs asso-
ciated with radiation-induced medical problems. These
results held robust throughout 1-way sensitivity analyses
and a probabilistic sensitivity analysis, in which proton
therapy was a cost-effective strategy compared with
photon therapy. For the 2 scenarios in which proton ther-
apy did not dominate in 1-way analysis, its ICER fell at
values less than $5000, which is well below norms for
what is considered a societal WTP threshold in the
United States.35 Furthermore, the probabilistic sensitivity
analysis revealed that the ICER robustly illustrated the
domination of photon therapy by proton therapy in
>95% of its 1000 iterations.
A notable strength of this work is its incorporation
of multiple, relevant health conditions in adulthood: 10
distinct health states were modeled for patient entry. Also,
costs and benefits associated with the health states were
informed after an extensive review of the literature using
the most recent data available on the risks and costs associ-
ated with both therapies. Furthermore, for this study, we
used primary data on the cost of photon and proton capi-
tal investment. Throughout the model, the costing strat-
egies used were conservative: the management of CAD
and CHF did not incorporate high expenses, such as
emergency room visits, hospitalization, or surgical inter-
vention, which may occur throughout the natural history
of disease. Instead, these strategies were modeled assum-
ing complete pharmacologic management. Similarly, the
proton strategy dominated the photon arm despite the
exclusion of pediatric costs, such as special education and
vocational rehabilitation associated with recently diag-
nosed hearing loss, which have lifetime cost estimates of
$400,000 and $12,000, respectively.26 These conservative
approaches avoided overestimating health state costs and,
thus, favored analysis findings toward acceptance of the
photon strategy, because these events would be associated
with the photon strategy with greater frequency. Conse-
quently, findings indicating the cost effectiveness and
domination of protons are more significant.
Many who argue in favor of developing proton
facilities cite the domination of proton therapy over
standard photon therapy in the management of childhood
cancer. However, since the original publication by
Lundkvist and colleagues, several assumptions may no
longer hold true, and more outcomes and modeling data
have been published, allowing for more precise ICER esti-
mates.7,13 The former model used the same utilities to

Cancer December 15, 2013 4303

Original Article

TABLE 4. One-Way Sensitivity Analysesa

Parameter Change D EFF, QALYs D Cost, $ ICER, $/QALY

Base case 3.46 232,579.08 Proton dominates

Cost of second cancer, $13,005.49b 3.48 230,255.36 Proton dominates
Proton-to-photon cost ratioc
32.4 3.46 29931.65 Proton dominates
33.2 3.47 17,014.13 4910
Risk of second cancer 2.71 229,000.09 Proton dominates
Risk of CAD 3.22 233,542.37 Proton dominates
Risk of CHF 2.65 233,705.22 Proton dominates
Risk of GHD 3.29 216,095.94 Proton dominates
Risk of hypothyroidism 2.93 232,040.24 Proton dominates
Risk of gonadotropin deficiency 3.10 227,834.37 Proton dominates
Risk of hearing loss 1.75 223,042.94 Proton dominates
Risk of ACTH deficiency 3.36 232,259.05 Proton dominates
Omission of ACTH, gonadotropin deficiency, and GHD 3.45 8663.65 2510

Abbreviations: ACTH, adrenocorticotropic hormone; CAD, coronary artery disease; CHF, congestive heart failure; EFF, effectiveness; GHD, growth hormone
deficiency; ICER, incremental cost-effectiveness ratio; QALY, quality-adjusted life year.
a
These analyses were run with 100,000 patients. For adverse event risk testing, the risk used for protons was modified to equal the risk for photons in the
base case: e.g., to test hearing loss, the proton arm value was set to 0.62, the photon arm value.
b
This value represents the estimated cost of cancer at 2 standard deviations below the average value.
c
Ratios were chosen from costing studies by Goitein & Jermann 200330 and Peeters 20103 for proton and photon therapy, respectively.

Figure 3. A probabilistic sensitivity analysis is illustrated. This chart displays the results from 1000 separate simulations of 25,000
patients. In each of those separate simulations, selected risks and costs varied according to the set distributions chosen to reflect
the uncertainty for those parameters. Each point represents 1 of those simulations and is charted at the simulation’s resultant
incremental cost versus incremental effectiveness of proton therapy compared with photon therapy. WTP indicates willingness
to pay.

reflect the preferences of participants in pediatric and the manipulation of adult utilities for children may inap-
adult years of life. However, pediatric utility use for propriately reflect health effects.36 Our study provided a
QALY determination is a frequently contested issue, and novel solution to studies involving pediatric illness by

4304 Cancer December 15, 2013


focusing on adult survivors. Therefore, the model allowed
for the incorporation of adult utilities for proper prefer-
ence selection; and, because of the lifetime duration of the
model, enough time was allowed to elapse to capture
meaningful differences in QALYs gained and costs
incurred. Compared with prior work, the current study
expanded the health states that were included in the model
to reflect a more accurate depiction of sequelae after treat-
ment with the inclusion of ACTH deficiency, gonadotro-
pin deficiency, and CAD. To our knowledge, this study is
also the first CEA of pediatric medulloblastoma to esti-
mate capital and investment costs from primary data, thus
not only allowing for increased precision of photon costs
but, even more important, allowing for a true representa-
tion of proton costs, which were further validated com-
pared with the costs to society obtained by Peeters et al
and Goitein and Jermann.3,37 Goitein and Jermann pre-
dicted a decrease in the proton:photon cost ratio with
time, and our study’s estimate corroborates that analysis.
Further costing accuracy was obtained by including not
only pharmacologic management costs, which had been
done in former studies, but also monitoring and evalua-
tion costs. A former review by Pijls-Johannesma and col-
leagues commented on the lack of a PSA in cost-
effectiveness studies that evaluated proton therapy.4 To
our knowledge, this is the first CEA of pediatric medullo-
blastoma to incorporate a PSA in its sensitivity analysis.
It is noteworthy that, despite the incorporation of
recently published data, the current study was still limited
by the paucity of data available on the clinical outcomes of
these patients. Premature ovarian failure, which may
result from cerebrospinal irradiation, could not be
included secondary to the lack of risk estimates in such
patients. Early puberty was not included despite evidence
from cohort studies, because its effects do not extend past
adolescence and would not be captured by the model.
Even so, these omitted health states would most likely
incur greater average costs for photon therapy patients
than for proton therapy patients because of the dose-
sparing properties of proton technology. Important
assumptions made were the independence of adverse
events and the constant rate of disease incidence for CHF,
CAD, and SMN, which may not reflect the true nature of
disease. Although both GHD and ACTH deficiency can
impact mortality when not properly managed, for this
study, we assumed that, apart from CAD, CHF, and
SMN, health states did not affect death. This decision was
based on the assumption of perfect detection, manage-
ment, and compliance with medication. Another limiting
assumption regarded the health state of gonadotropin
Cancer December 15, 2013
CEA of Proton vs Photon RT for Medullo/Mailhot Vega et al
deficiency, which is known to affect men and women dis-
tinctly for both management and utility. However, there
is no evidence suggesting that men or women have differ-
ent susceptibilities to developing gonadotropin defi-
ciency. Hence, although the study was limited in that
both utilities and costs were averaged, mathematically, the
analysis should be unaffected. Compared with prior work,
the most notable limitation of this study was its omission
of cognitive impairment, the most influential parameter
in the former CEA for pediatric medulloblastoma by
Lundkvist and colleagues.12 It is probable that there is a
greater loss of productivity from cognitive decline in
patients who receive photon therapy; however, the use by
Lundkvist et al of the cost to society per decrease in IQ
score was derived from lead toxicity ecologic studies.
Thus, this cost may not accurately reflect the cost of IQ
loss in the setting of cranial irradiation owing to differen-
tial mechanisms of neurotoxicity and neurobehavioral
outcomes.38-40 To date, there have been no longitudinal
studies detailing the cost of productivity loss secondary to
cognitive decline in childhood cancer survivors managed
with cranial irradiation, and that absence prompted our
omission of cognitive decline as a health state to avoid
inappropriate cost assignment for IQ loss. However, we
can provide an estimate of the magnitude of cognitive
impairment’s effect on the predicted cost effectiveness of
proton therapy. Similar to Lundkvist et al, we can assume
that a loss of productivity secondary to cognitive impair-
ment exists, and a subsequent cost can be estimated by the
differential IQ decrement’s effect on income. We can
assess this cost with 2 separate methodologies to provide a
range: an approach used by Schwartz and colleagues (ref-
erenced by Lundkvist et al) and 1 used by Zagorsky and
colleagues.12,41,42 Secondary to the 10-point IQ differ-
ence projected between groups by Merchant and col-
leagues,43 there exists a differential annual wage reduction
of 5% in the photon group according to Schwartz et al
and of $5000 according to Zagorsky et al. Assuming a
child aged 5 years today will participate in the work force
from age 18 years until age 65 years, a 2% cost-of-living
adjustment per year, and a 3% discounting per year, the
projected cost of productivity loss is approximately
$77,482 and $169,636, respectively.40
The objective of this study was to assess whether
proton therapy is a cost-effective strategy compared with
photon therapy in the management of pediatric medullo-
blastoma. In this study, proton therapy dominated pho-
ton therapy in the base case, but these results may differ
with more accurate long-term follow-up of treated
patients on current protocols. The most accurate way to
4305
Original Article
determine comparative effectiveness would be a phase 3
trial comparing photons with protons in an adequately
sized population; however, given the apparent dosimetric
differences and the known late effects of radiation, that
kind of study would be considered unethical by many.
This study did not answer whether the cost effectiveness
of protons in pediatric medulloblastoma translated to cost
effectiveness in other malignancies and did not justify the
expense of more proton facilities. The current study also
did not address the cost effectiveness of including strat-
egies with intensity modulation, such as intensity-
modulated radiation therapy and intensity-modulated
proton therapy, because even less clinical data are available
for these modalities. Although proton therapy may domi-
nate photon therapy in pediatric medulloblastoma, the
prevalence of the disease is low and may not provide suffi-
cient patient volume for the cost-effective use of many
facilities. Future studies could elucidate the range of
acceptability of proton therapy compared with photon
therapy across a spectrum of malignancies, and proper
phase 3 trials are needed to provide necessary risk values
to inform models.
FUNDING SUPPORT
Dr. Mailhot Vega was supported by a medical student fellowship
from the American Society for Therapeutic Radiology and Oncol-
ogy. The project was in part supported by the Federal Share of pro-
gram income earned by Massachusetts General Hospital on C06
CA059267, Proton Therapy Research and Treatment Center.
CONFLICT OF INTEREST DISCLOSURES
Dr. Tarbell’s spouse serves on the medical advisory board of
Procure.
REFERENCES
1. Delaney TF. Proton therapy in the clinic. Front Radiat Ther Oncol.
2011;43:465-485.
2. Sher DJ. Cost-effectiveness studies in radiation therapy. Expert Rev
Pharmacoecon Outcomes Res. 2010;10:567-582.
3. Peeters A, Grutters JP, Pijls-Johannesma M, et al. How costly is
particle therapy? Radiother Oncol. 2010;95:45-53.
4. Pijls-Johannesma M, Pommier P, Lievens Y. Cost-effectiveness of
particle therapy. Radiother Oncol. 2008;89:127-134.
5. Epstein K. Is spending on proton beam therapy for cancer going too
far, too fast [serial online]? BMJ. 2012;344:e2488.
6. Levin WP, Kooy H, Loeffler JS, DeLaney TF. Proton beam therapy.
Br J Cancer. 2005;93:849-854.
7. Brodin NP, Munck Af Rosenschold P, Aznar MC, et al. Radiobio-
logical risk estimates of adverse events and secondary cancer for
proton and photon radiation therapy of pediatric medulloblastoma.
Acta Oncol. 2011;50:806-816.
8. Freeman CR, Taylor RE, Kortmann RD, Carrie C. Radiotherapy
for medulloblastoma in children: a perspective on current interna-
tional clinical research efforts. Med Pediatr Oncol. 2002;39:99-108.
9. Bailey CC, Gnekow A, Wellek S, et al. Prospective randomised trial
of chemotherapy given before radiotherapy in childhood medullo-
blastoma. International Society of Paediatric Oncology (SIOP) and
4306
the (German) Society of Paediatric Oncology (GPO): SIOP II. Med
Pediatr Oncol. 1995;25:166-178.
10. Fossati P, Ricardi U, Orecchia R. Pediatric medulloblastoma: toxic-
ity of current treatment and potential role of proton therapy. Cancer
Treat Rev. 2009;35:79-96.
11. St. Jude Children’s Research Hospital. Childhood Cancer Survivor
Study. Hormonal systems: all cancers. Memphis, TN: St. Jude
Children’s Research Hospital; 2012 [updated September 15, 2008].
Available at: http:==ccss.stjude.org=docs=ccss=data-bk-hormonal-sys-
baseE.pdf. Accessed August 1, 2012.
12. Lundkvist J, Ekman M, Ericsson SR, Jonsson B, Glimelius B. Cost
effectiveness of proton radiation in the treatment of childhood
medulloblastoma. Cancer. 2005;103:793-801.
13. Yock TI, Yeap BY, Ebb D, et al. A phase II trial of photon radio-
therapy for medulloblastoma: preliminary results [abstract]. J Clin
Oncol. 2010;28(suppl 18). Abstract CRA9507.
14. Arias E. United States life tables, 2007. Natl Vital Stat Rep. 2011;
59:1-60.
15. Merchant TE, Rose SR, Bosley C, Wu S, Xiong X, Lustig RH.
Growth hormone secretion after conformal radiation therapy in pe-
diatric patients with localized brain tumors. J Clin Oncol. 2011;29:
4776-4780.
16. Frange P, Alapetite C, Gaboriaud G, et al. From childhood to adult-
hood: long-term outcome of medulloblastoma patients. The Institut
Curie experience (1980-2000). J Neurooncol. 2009;95:271-279.
17. Heikens J, Michiels EM, Behrendt H, Endert E, Bakker PJ, Fliers E.
Long-term neuro-endocrine sequelae after treatment for childhood
medulloblastoma. Eur J Cancer. 1998;34:1592-1597.
18. Miralbell R, Lomax A, Russo M. Potential role of proton therapy in
the treatment of pediatric medulloblastoma=primitive neuro-
ectodermal tumors: spinal theca irradiation. Int J Radiat Oncol Biol
Phys. 1997;38:805-811.
19. Sullivan PW, Ghushchyan V. Preference-Based EQ-5D index scores
for chronic conditions in the United States. Med Decis Making.
2006;26:410-420.
20. Ekman B, Bachrach-Lindstrom M, Lindstrom T, et al. A random-
ized, double-blind, crossover study comparing 2- and 4-dose hydro-
cortisone regimen with regard to quality of life, cortisol and ACTH
profiles in patients with primary adrenal insufficiency. Clin Endocri-
nol (Oxf). 2012;77:18-25.
21. Koltowska-Haggstrom M, Mattsson AF, Shalet SM. Assessment of
quality of life in adult patients with GH deficiency. Eur J Endocri-
nol. 2009;161:S51-S64.
22. Krahn M, Ritvo P, Irvine J, et al. Patient and community preferen-
ces for outcomes in prostate cancer. Med Care. 2003;41:153-164.
23. Packer RJ, Gajjar A, Vezina G, et al. Phase III study of craniospinal
radiation therapy followed by adjuvant chemotherapy for newly diag-
nosed average-risk medulloblastoma. J Clin Oncol. 2006;24:4202-
4208.
24. Cook DM, Yuen KC, Biller BM, Kemp SF, Vance ML. American
Association of Clinical Endocrinologists medical guidelines for clini-
cal practice for growth hormone use in growth hormone-deficient
adults and transition patients—2009 update. Endocr Pract. 2009;
15(suppl 2):1-29.
25. Baskin HJ, Cobin RH, Duick DS, et al. American Association of
Clinical Endocrinologists medical guidelines for clinical practice for
the evaluation and treatment of hyperthyroidism and hypothyroid-
ism. Endocr Pract. 2002;8:457-469.
26. Mohr PE, Feldman JJ, Dunbar JL, et al. The societal costs of severe
to profound hearing loss in the United States. Int J Technol Assess
Health Care. 2000;16:1120-1135.
27. Grossman AB. The diagnosis and management of central hypoadren-
alism. J Clin Endocrinol Metab. 2010;95:4855-4863.
28. Snyder PJ. Testosterone treatment of male hypogonadism. In:
Basow, DS, ed. UpToDate. Waltham, MA: UpToDate; 2012.
29. Nelson L, Calis KA. Management of spontaneous primary ovarian
insufficiency. In: Basow, DS, ed. UpToDate. Waltham, MA: UpTo-
Date; 2012.
30. Hunt SA, Abraham WT, Chin MH, Feldman AM, Francis GS,
Ganiats TG. 2009 Focused update incorporated into the ACC=AHA
2005 Guidelines for the Diagnosis and Management of Heart
Cancer December 15, 2013

Failure in Adults: a report of the American College of Cardiology
Foundation=American Heart Association Task Force on Practice
Guidelines: developed in collaboration with the International Society
for Heart and Lung Transplantation [serial online]. Circulation.
2009;119:e391-e479.
31. Fraker TD Jr, Fihn SD, Gibbons RJ, et al. 2007 Chronic angina
focused update of the ACC=AHA 2002 guidelines for the manage-
ment of patients with chronic stable angina. J Am Coll Cardiol.
2007;50:2264-2274.
32. PDR Network. Red Book: Pharmacy’s Fundamental Reference.
Montvale, NJ: PDR Network; 2010.
33. Bureau of Labor Statistics. CPI Inflation Calculator. Washington,
DC: US Department of Labor, Bureau of Labor Statistics; 2012.
Available at: http:==www.bls.gov=data=inflation_calculator.htm.
Accessed August 1, 2012.
34. Russell LB, Gold MR, Siegel JE, Daniels N, Weinstein MC. The
role of cost-effectiveness analysis in health and medicine. Panel on
Cost-Effectiveness in Health and Medicine. JAMA. 1996;276:1172-
1177.
35. Eichler HG, Kong SX, Gerth WC, Mavros P, Jonsson B. Use of
cost-effective analysis in health-care resource allocation decision-mak-
ing: how are cost-effectiveness thresholds expected to emerge? Value
Health. 2004;7:518-528.
36. Ungar WJ. Challenges in health state valuation in paediatric eco-
nomic evaluation: are QALYs contraindicated? Pharmacoeconomics.
2011;29:641-652.
Cancer December 15, 2013
CEA of Proton vs Photon RT for Medullo/Mailhot Vega et al
37. Goitein M, Jermann M. The relative costs of proton and x-ray radia-
tion therapy. Clin Oncol (R Coll Radiol). 2003;15:S37-S50.
38. Armstrong GT, Liu Q, Yasui Y, et al. Long-term outcomes among
adult survivors of childhood central nervous system malignancies in
the Childhood Cancer Survivor Study. J Natl Cancer Inst. 2009;101:
946-958.
39. Bressler J, Kim KA, Chakraborti T, Goldstein G. Molecular mecha-
nisms of lead neurotoxicity. Neurochem Res. 1999;24:595-600.
40. Calabrese P, Schlegel U. Neurotoxicity of treatment. Recent Results
Cancer Res. 2009;171:165-174.
41. Schwartz J. Societal benefits of reducing lead exposure. Environ Res.
1994;66:105-124.
42. Zagorsky JL. Do you have to be smart to be rich? The impact of IQ
on wealth, income and financial distress. Intelligence. 2007;35:489-
501.
43. Merchant TE, Hua C, Shukla H, et al. Proton versus photon radio-
therapy for common pediatric brain tumors. Pediatr Blood Cancer.
2008;51:110-117.
44. US Bureau of Labor Statistics. National Compensation Survey:
Occupational Earning in the United States, 2010. Washington, DC:
United States Department of Labor, Bureau of Labor Statistics;
2010. Available at: http:==www.bls.gov=ncs=ocs=sp=nctb1477.pdf.
Accessed June 1, 2013.
4307