Clinical Oncology (2002) 14: 382393

doi:10.1053/clon.2002.0111, available online at http://www.idealibrary.com on

Original Article
Practical Methods for Compensating for Missed Treatment
Days in Radiotherapy, with Particular Reference to Head
and Neck Schedules
R. G. Dale*, J. H. Hendry, B. Jones*, A. G. Robertson, C. Deehan,
J. A. Sinclair*
*Hammersmith Hospitals NHS Trust/Imperial College School of Medicine, London, U.K.; Paterson Institute for
Cancer Research, Christie Hospital NHS Trust, Manchester, U.K.; Beatson Oncology Centre, Western Infirmary,
Glasgow, U.K.; Department of Medical Physics, Leicester Royal Infirmary, Leicester, U.K.
ABSTRACT:
Unscheduled interruption of a radiotherapy treatment can lead to significant loss in local tumour control, particularly in tumours
that repopulate rapidly. General guidelines for dealing with such treatment gaps have been issued by the Royal College of
Radiologists and more specific advice on the use of compensation methods has been published previously [Hendry et al., Clin Oncol
1996;8:297307; Slevin et al., Radiother Oncol 1992;24:215220]. This article further elaborates on the practical application of these
methods. It sets out the main considerations arising in the especially critical case of head and neck treatments and simple
calculations are used to illustrate the approaches which may be adapted for particular situations. Radiobiological parameter values
are suggested for use in the calculations, but these may require modification in the light of further research in this important area.
Dale, R. G. et al. (2002). Clinical Oncology 14, 382393
 2002 The Royal College of Radiologists. Published by Elsevier Science Ltd. All rights reserved.
Key words: Cancer, cell kinetics, fractionation, overall time, radiotherapy, time factors in radiotherapy, treatment interruptions
Received: 5 July 2001

Introduction

Local tumour control can be adversely aected by

prolongation of overall treatment time in radiotherapy.
A number of studies have highlighted that tumour cell
repopulation can produce losses in local control which,
particularly for squamous cell cancers of the head and
neck region, may amount to 12% per day of treatment
prolongation [1,2]. Similarly, recent studies on nonsmall cell lung cancer and cervix carcinoma have indicated respectively 1.6% and 0.8% loss of local control
per day of treatment extension [3,4]. Adherence to
radiotherapy schedules is therefore important if individual treatments are not to be compromised. However,
there are occasions when unforeseen interruptions
occur, prime causes being equipment breakdown or
unwell patients and the Royal College of Radiologists
(RCR) in 1996 issued guidelines for dealing with
unscheduled treatment interruptions [5]. In parallel with
these guidelines a more detailed review of the evidence
for the deleterious eects of treatment prolongation was
Author for correspondence: Dr R. G. Dale, Radiation, Physics &
Radiobiology, Charing Cross Hospital, London W6 8RF, U.K.
E-mail: r.dale@ic.ac.uk
09366555/02/040382+12 $35.00/0

Accepted: 2 May 2002

published, together with a comparative assessment of

the methods available for the compensation of missed
treatment days [1].
Many U.K. departments have since implemented
measures for handling unscheduled treatment interruptions and the evolving experience has highlighted that
some diculties remain, particularly with respect to
terminology and in relation to how to use the various
radiobiological parameter values. This article reiterates
some of the points previously discussed [1], but in
addition it specifically elaborates on the practical issues
which arise when devising compensations for interrupted treatments. Although some methods of treatment
compensation are clearly preferable to others, practical
factors (machine availability, ability of centres to work
extended hours and/or weekends, ability of an unwell
patient to undergo the full compensation, etc.) also
influence the decision regarding which form of compensation to adopt. It is therefore of some importance
(particularly in more dicult cases) to examine two or
three possible solutions in order to identify a compromise scheme which helps the patient without causing
severe disruption to local service arrangements. The
methods described below are those to be considered

 2002 The Royal College of Radiologists. Published by Elsevier Science Ltd. All rights reserved.

383

Fig. 1 Schematic showing how the total dose required for a given tumour control probability (TCP) changes with increasing
treatment duration. The horizontal and sloped lines demonstrate how the dose rises steadily after an initial delay period.

when technical failures have caused the interruption.

For gaps resulting from adverse clinical reactions the
methodology is the same but the condition of the patient
may mean that it is not always possible to compensate to
the same extent. Although this paper concentrates on
head and neck squamous cell cancer, the methodology
is applicable to other sites provided appropriate
alternative parameter values are used.

Influence of unscheduled treatment interruptions on tumour

repopulation

Figure 1 is a simplified diagram showing how the total

(physical) dose for a given tumour response (e.g. 50%
tumour control probability when the dose is delivered in
2 Gy fractions) is assumed to vary with treatment time.
If the treatment is completed before the initiation of
accelerated repopulation (denoted by the rising line
beginning at time Tdelay) then the dose required in this
region is fixed and is independent of the treatment time
used. If, however, the treatment is of duration greater
than Tdelay days, the iso-eect dose increases in order to
compensate for the consequent tumour repopulation,
the dose in this region now being time-dependent. The
dose prescribed in schedules which extend beyond Tdelay
days will inherently include extra dose to compensate for
the repopulation eect.
Figure 2 illustrates a prescribed treatment which is
scheduled to be completed at time T. If there occurs an
unscheduled interruption of duration t days during the
treatment, and no allowance is immediately made for
this, then the treatment will need to be extended by
t days at the end of the prescribed schedule. To maintain
the tumour iso-eect would require the addition of
extra dose, incurring an associated penalty (impaired
Therapeutic Index) in terms of a higher normal tissue
dose. To prevent overdosing of the normal tissue would

require that the supplementary tumour dose be reduced,

also with detrimental eect.
If, however, the unscheduled interruption could be
compensated for by delivering the missing dose (by
extra daily fractions, or weekend treatments), without
extending thetreatment duration, no such penalty would
be incurred. This demonstrates an important point: it is
the likelihood of extension to the treatment time, rather
than the gap per se, which is the main cause for concern
when interruptions occur. A further point is that
unscheduled gaps occurring early in treatment (before
the onset of tumour repopulation), if they lead to a
treatment extension, are assumed to be no less detrimental (on current evidence) than those occurring later in
treatment. The dierence between an early- and lateinterruption is practical rather than radiobiological:
with the former more time remains within the prescribed
schedule to eect a satisfactory compensation.
More detailed schematics illustrating the problems of
treatment interruptions have been published elsewhere
[68].
Equations

The biologically eective dose (BED) received by

a uniformly irradiated tissue which is concurrently
repopulating is calculated as:

where n is the number of fractions, d is the dose per

fraction, T is the overall treatment time. Tdelay is the
delay time (from the beginning of treatment) before the
onset of significant repopulation [9].
/ (in units of Gy)
is the fractionation factor and K (in units of Gy day
1)
is the biological dose per day required to compensate for

384

Fig. 2 Showing the eect of a treatment extension, caused by an earlier unscheduled gap. If the treatment extends into the time
period where steady tumour repopulation is occurring, extra dose (X) is required to compensate for this. It should be noted that this
will be the case even if, as here, the unscheduled gap occurred during the period prior to the initiation of tumour repopulation.

on-going tumour cell repopulation. K is the BEDequivalent of 1 days worth of repopulation [9]. The
expression in square brackets in Eqn (1) is termed the
Relative Eectiveness factor (RE).
Both
/ and K are tissue-specific and appropriate
values must be selected for each when calculating
tumour BEDs.
As applied to a tumour, Eqn (1) may be remembered
in words as:
BED=Total physical dose RERepopulation Factor
(RF)
For most late-responding normal tissues the proliferation rate is usually so small (except in cases where
consequential late reactions are dose-limiting [10])
that K can be neglected, i.e. K=0 and RF=0 in such
cases.
It has become conventional practice to place the
/
value used in the BED calculation as a subscript to both,
the BED symbol and its associated numerical value. For
example, a stated BED3 of 100 Gy3 indicates that

/=3 Gy was used in calculating that particular biological dose. The use of the subscript reinforces the point
that biological dose is conceptually dierent to physical
dose, even though both are in similar dimensions. Biological doses expressed in (for example) Gy10 can be
added to other Gy10 values to provide a measure
of resultant eect, but it is not permissible to add (say)
Gy3 values to Gy10 values.
Eqn (1) is valid in cases where the fractions are
relatively well-spaced. When two or more fractions are
delivered per day there may be incomplete repair of

sub-lethal damage between successive fractions, leading

to increased biological damage [11]. This is particularly
so withregard to late reactions. In such cases Eqn (1) is
changed to:

Factor h in Eqn (4) reflects the increased damage caused

by incomplete repair. It is complex in form and is
dependent on the inter-fraction intervals (which are not
usually the same between all fractions) and the repair
kinetics. The derivation of the h factor is discussed in the
Appendix. There may be occasions when compensation
for unscheduled treatment gaps will involve the use of
many relatively close fractions (less than 8 h between
fractions) and, in such cases, the eects of incomplete
repair should be borne in mind. If there is to be more
than one day on which twice-or thrice daily fractionation is required then, where possible, such days should
not be consecutive. It is recommended that allowances
for incomplete-repair should always be included when
the use of four or more closely-spaced fractions is
unavoidable. A later example (Example 5) will illustrate
some of the issues involved.
Eqns (1) and (2) provide the basis for devising compensation for unscheduled treatment interruptions, but
there are many instances where they need to be utilized
in a slightly dierent form. Rather than list a family of
related equations, the methodology appropriate to particular circumstances will be explained in the worked
examples.


Recommended values of
/, K and Tdelay

The range of
/ values usually considered for head and

neck cancers is around 1020 Gy [12]. A preliminary
analysis of the recently reported RTOG9003 head and
neck trial indicates a value just below the bottom end of
this range [13]. We therefore suggest that a generic
/
value of 10 Gy be used for head and neck cancers, but
this figure should be kept under review.
For late-responding normal tissues a generic
/ of
3 Gy is generally recommended. The important exceptions to this are brain and spinal cord, for each of which
an
/ value of 2 Gy should be used, reduced further
to 1.5 Gy when fraction sizes greater than 2 Gy are
employed. A reduced
/ of 2.5 Gy may also be preferable for normal tissues in cases where there are concerns
that tolerance may be clinically compromised [11]. A
comprehensive tabulation of
/ values for various normal tissue and tumours can be found in the chapter by
Joiner and van der Kogel in reference [14].
The data reviewed for the previous Clinical Oncology
article on gap compensation [1] indicated that head and
neck K factors were in the range 0.50.74 Gy day
1. Since
then more sophisticated analyses of multi-centre data have
shown that K values are likely to be higher than originally
believed. Roberts and Hendry [15] have conducted a metaanalysis of larynx treatment results from Edinburgh,
Glasgow, Manchester and Toronto, leading to an upwardly
revised K estimate of 1.2 (95% CL 0.82.2) Gy day
1. The
results of the RTOG9003 head and neck trial [16] also
lend support to the likelihood of higher K values with
early analyses of those data suggesting K values of
0.940.99 Gy day
1 [13,17].
The head and neck tumour control data collated by
Withers et al. [12] demonstrated a dog-leg eect of
the type illustrated in Fig. 1 and have been analysed in
detail by Hendry and Roberts [18], who found the
repopulation delay time (Tdelay) to be 29 (95% CL
1731) days. For the shorter Manchester treatment
regimes the delay time was established as 26 (95% CL
1933) days.
On the basis of the above, and bearing in mind the
relatively large confidence intervals and the fact that the
fitted values of K and Tdelay must be used together, we
suggest that respective working values of 0.9 Gy day
1
and 28 days be adopted for head and neck cancers. It is
stressed that neither of these recommended values can
yet be considered as definitive and that they must be
reviewed in the light of continuing research.
The recommended K value of 0.9 Gy day
1 represents the BED required each day (after Tdelay has been
passed) to oset repopulation. If
/ for tumours is
taken as 10 Gy then, more specifically, K is the BEDequivalent of repopulation in units of Gy10 per day. This
means that, when 2 Gy fractions are being used, the
daily physical dose required to oset the repopulation
after time Tdelay is K/[1+2/10]=0.75 Gy day
1, i.e. for
each 2 Gy fraction delivered, (0.75/2)100=38% is
wasted in combating repopulation.

385

Values of K and Tdelay are derived on the basis that

there is an intrinsic pairing between them. This arises
because there is an a priori assumption that there is zero
population in the period up to Tdelay, meaning that the
paired parameter values provide the best fit to the
observed increases in dose over the whole range of
treatment times studied.
The data relating to time factors and the eects of
gaps for head and neck treatments were reviewed up to
1996 in the previous article [1]. That review discussed the
fact that the time factor was approximately the same
during protracted treatments as in split-course treatments and that there were contradictory suggestions
regarding the relative importance of gaps early or late in
the schedule. Since 1996 the importance of treatment
interruptions in radiotherapy for oropharyngeal carcinomas has been highlighted [19]. Evidence has also
emerged that the position of the gap is not important
and that the loss of tumour control following each day
of interruption of a head and neck treatment is about
the same as occurs when conventional schedules are
protracted [20,21]. The latter feature was confirmed in a
human xenograft tumour model in mice [22]. In contrast, in rat tumours it has been shown that the position
of the gap does matter [23]. Regarding the combination
of chemotherapy and radiotherapy, there are reports of
a lack of a time factor in alternating chemoradiotherapy
for advanced head and neck squamous cell carcinoma
[24]. Long gaps were found to be detrimental to local
control in anal canal carcinoma treated by split-course
radiotherapy and concomitant chemotherapy [25].
As discussed earlier, the assumptions in the head and
neck data fitting exercises inherently excluded the possibility of there being any repopulation in the period up to
the Tdelay time point, but that does not mean that
repopulation does not occur in that time. Indeed, other
analyses suggest that repopulation does occur from the
beginning of treatment [26]. There thus arises the question of what K factor should be used in interrupted
treatments for which the final overall time is less than
Tdelay. Withers et al. [12] suggested that the repopulation
rate during the early part of treatment reflected the
pre-treatment tumour growth-rate. Examination of the
slopes before and after the delay time suggests they are
in the approximate ratio 1:10, i.e. that the K value to use
prior to Tdelay is one-tenth that used after the delay
point. For head and neck treatments we therefore recommend K=0.1 Gy day
1 and Tdelay =0 in cases where
the final overall times are less that 28 days. It is
emphasized that these values are suggested on the basis
of the evidence currently available; they cannot be
regarded as being definitive.
Although the methodology described in this paper is
more widely applicable, there is at present very little data
relating to K and Tdelay factors for other tumour types.
For non-small-cell lung tumours there is evidence that
the loss of local control per day towards the end of
treatment is about the same as that for head and neck
cancers [3,27]. For cervix tumours the time factors are

386

calculated via Eqn (1) but, for the late-reacting

normal tissue, K is set to zero.
(2) Determine the overall time until the beginning of
the unscheduled gap and, hence, the respective
pre-gap BEDs.
(3) Determine the late-normal BED still to give (the
post-gap BED).
(4) Review the various treatment options (e.g.
twice-daily fractionation and hyperfractionation,
increased fraction sizes, etc.) to ascertain which will
be likely to produce the minimum extension to the
treatment time, then calculate the required dose per

probably around half those for head and neck tumours,

i.e. c0.5 Gy day
1 [1]. For breast tumours the K factors are likely to be around 0.3 Gy day
1 while for
prostate tumours they are likely to be in the region
0.10.3 Gy day
1, or even lower [2830]. Little is
known about the corresponding values of Tdelay.
Calculation process

The table below (based on reference [1]) identifies the

main methods for compensation and the associated
benefits and diculties.

Method

Benefit

Diculty

(1a) Retain overall time and dose per fraction

by treating on weekend days as necessary.

Overall time, fraction size, inter-fraction

interval and Therapeutic Index maintained.

Unsocial hours and associated costs. May not

be appropriate for gaps occurring near the
end of a schedule.

(1b) Retain overall time and dose per fraction

by treating twice-daily as necessary.

Overall time and fraction size maintained.

Possible loss of late-normal tissue tolerance

when several fractions have to be delivered at
68 h inter-fraction intervals. Scheduling
diculties.

(2a) Retain overall time by increasing dose

per fraction for same number of post-gap
days as there were gap days.

Overall time retained by accepting reduced

number of fractions. Still one fraction on each
treatment day.

Not suitable for schedules which already use

high dose per fraction. Therapeutic Index
adversely aected: Seeking equivalence for
tumour control gives increase in late
reactions. Seeking equivalence for
late-reactions leads to tumour under-dosage.

(2b) Retain overall time by using smaller

number of larger fractions after the gap.

Overall time retained. Still one fraction per

day.

As above.

(3a) Accept that treatment extension is

unavoidable and deliver extra fractions, using
increased dose per fraction to minimize the
extension duration.

Allows at least partial restoration of the

prescribed schedule.

Therapeutic Index adversely aected. Might

require acceptance of both reduced tumour
control and increased late-eects.

(3b) As for 3(a) but use twice-daily fractions

and a slightly reduced treatment extension.

As above.

As for 3(a), but deterioration in Therapeutic

Index is not so marked.

Interrupted treatments need to be evaluated on an

individual basis and there is no universal method for
tackling all problems. However, most examples will
involve a series of general steps which may be
summarized as follows:
Once an unscheduled gap has occurred, first determine
the remaining treatment time and the number of fractions still to be delivered. Determine if there are ways of
delivering these fractions to allow the originally prescribed treatment time to be maintained, e.g. by treating
at weekends or by giving all or part of the remaining
treatment twice-daily. If this is possible then further
radiobiological calculations may not be necessary. (Examples 1 and 2 below relate to such a case). If this
option is not feasible (i.e. it is not possible to complete
treatment within the prescribed treatment time) then:
(1) First calculate the tumour and normal tissue
BEDs for the prescribed schedule. Both BEDs are

fraction to achieve the required late-normal BED

value. Calculate the resulting tumour BED for this
option, remembering to make allowance for the
extended time (Example 36 below demonstrate
various versions of this scenario).
(5) Review the final tumour and normal tissue BEDs
which will result from the preferred compensation
option. If the tumour BED is significantly smaller
than that originally prescribed a degree of clinical
judgement may be required in order to fine tune
the compensation in order to arrive at a reasonable
compromise. (Examples 46 illustrate the dilemmas
which may be involved in such cases).
It is stressed that these are general steps. For example, if
the favoured compensation option involves closelyspaced fractions after the gap, the modified BED formula [Eqn (2)] must be used to determine the possible
enhancement to normal tissue toxicity as a consequence
of incomplete-repair. It is suggested that if a total of four


fractions or more are to be given at twice-daily intervals
(or more frequently) then the eects of incomplete repair
should always be considered.
Worked examples

Worked examples 13 each consider the handling of

5-day gaps. In practice the majority of unscheduled
interruptions involve interruptions of 5 days or less and
are relatively easy to deal with. With a gap of just 1 or
2 days dierent policies may apply in dierent centres.
Examples 15 involve a reference schedule of 70 Gy
delivered in 35 fractions over 46 days, typically used for
Category 1 head and neck tumours. The overall time of
46 days corresponds to a treatment beginning on a
Monday, continues with daily-fractionation for 7 weeks
with no treatment at weekends and finishes on a Friday.
For a similar 35 fraction schedule which begins midweek the treatment time will be longer (because the
treatment will extend into an eighth week) and specific
calculations should allow for this.
For other schedules, e.g. the commonly used 4-week
treatments, the principle involved in determining a
method of compensation is exactly the same as set out in
the 7-week treatments used here. In such cases, however,
there is more concern about b.i.d treatments if the dose
per fraction is already significantly larger than 2 Gy,
because of the greater potential for incomplete repair.
Example 6 elaborates on this and also discusses a
treatment which does not begin on a Monday.
In all the examples requiring calculations, the process
here has been to devise a compensation scheme to
maintain the desired late-reacting BED and then, as
necessary, to consider alternatives in the light of how
much the resultant tumour BED is compromised. This is
done because late morbidity is often the most critical
concern in radiotherapy and some normal tissue parameter values are more reliably known than those for
tumours. Also, because late-responding tissues are
associated with less heterogeneity than tumours they
often exhibit steeper dose-response curves, i.e. any
deviations from the BEDs associated with the original
schedule will be more critical for the normal tissues.
Notwithstanding this view, however, there may be
instances where clinicians decide from the outset to
accept the risk of an elevated late-normal BED in order
to maintain a desired tumour eect.
Example 1. Loss of all of the third week (five fractions) of a
treatment schedule of 70 Gy/35 fractions/46 days.

Assuming the treatment began on a Monday, the

intended overall treatment time is 46 days. After the gap,
treatment resumes on the Monday of the fourth week of
the schedule. Ten fractions have been delivered; 25
remain to be given. If treatment is to be completed on
the prescribed finishing date the available number of
days (including weekends) is 26. Thus the missed dose
in the gap can be compensated for by delivering the

387

remainder of the treatment on weekdays (20 fractions)

and on five of the six remaining weekend days. This
does not involve changing the fraction size and, as
the treatment is not extended, constitutes a good
compensation.
If weekend treatments are not feasible a good compensation is still possible if, on five of the 20 remaining
treatment days, two fractions are delivered instead of
one. The important proviso is that the twice-daily fractions must be delivered with a minimum time gap
between them of at least 6 h, preferably 8 h, in order to
minimize any potential problems with incomplete repair
[31]. It is further recommended that the days on which
twice-daily treatments are delivered are not consecutive,
but spaced throughout the available time period. In this
instance Fridays are a good choice for delivery of some
of the twice-daily fractions as there is a greater opportunity for completion of repair before treatment resumes
the following week.
Example 2. Loss of all of the sixth week (five fractions) of a
treatment schedule of 70 Gy/35 fractions/46 days.

After the gap, treatment resumes on the Monday of the

seventh week of the schedule. Twenty-five fractions have
been delivered; 10 remain to be given. Ideally these 10
fractions should be delivered over the five remaining
treatment days. The missed dose can therefore be
compensated for by delivering the remainder of the
treatment as twice-daily fractions in each weekday
of the final week. This does not involve changing the
fraction size and, as the treatment is not extended,
constitutes a good compensation. A better solution, if
feasible, would be also to make use of the weekend
before the final week of treatment, thus providing 7 days
within which 10 fractions have to be delivered. Bi-daily
fractionation could be used (for example) on Monday,
Wednesday and Friday, single fractions on the other
4 days. The advantage of the latter scheme is that
it reduces the likelihood of creating excess biological
damage if there is incomplete repair between fractions.
Examples 1 and 2 represent sound solutions for
dealing with unscheduled interruptions; they do not
involve changing fraction size or overall time and,
provided there is reasonable spacing between treatment
days on which bi-daily treatment is given, do not invoke
any quantitative evaluations or serious radiobiological dilemmas. The following examples illustrate the
compromises involved in more dicult cases.
Example 3. Loss of all of the seventh week (five fractions)
of a treatment schedule of 70 Gy/35 fractions/46 days.

For the prescribed treatment the normal tissue BED

(BED3) is, from Eqn (1) with K=0:

388

The tumour BED (BED10), also from Eqn (1) but with
K=0.9 and Tdelay =28 days is:

In this example the unscheduled gap extends to the time

when treatment should have finished and any form of
compensation will therefore extend the treatment time
beyond the scheduled time.
We begin by assuming that the missing dose is
replaced by treating five 2 Gy fractions over a full extra
(eighth) week, beginning on a Monday. On completion,
the overall time is 7 days longer than scheduled. With
a daily BED-equivalent of tumour repopulation of
0.9 Gy day
1 the tumour BED10 will be lower than
intended by an amount 70.9=6.3 Gy10. The latenormal BED3 will be as prescribed.
If, instead, the outstanding daily treatments are given
in the period SaturdayWednesday, the net treatment
extension is 5 days, i.e. the tumour BED10 is low by a
smaller factor of 50.9=4.5 Gy10. A further alternative
is to treat two fractions per day on Saturday and
Monday with one fraction on Sunday, thus extending
treatment by 3 days. In this case the tumour BED10 will
be low by an even smaller amount of 30.9=2.7 Gy10.
In each of these instances the normal tissue BED3 will be
as prescribed.
The dilemmas arise when attempts are made to
increase the total dose in order to restore the tumour
BED10 to that originally prescribed; in this case it is
impossible to do that without increasing the normal
tissue BED3 [8]. Delivering extra dose by treating with
extra fractions has the eect of further extending the
treatment time, which may compound the original problem. Increasing the dose per fraction helps minimize the
treatment extension but, because of the greater sensitivity of the late-responding critical tissue to changes in
dose per fractions, will increase the normal tissue biological dose proportionately more than that for the
tumour.
We next consider an instance where it is felt essential
to restore the tumour BED10 to what it should be,
initially without regard for the eect on the normal
tissue. We assume the option of treating additionally
over the weekend is to be adopted, i.e. the overall time is
46+5=51 days.
The tumour BED10 of 67.8 Gy10 is to be maintained.
Therefore, for the whole schedule (pre-gap plus postgap):

where d is the new value of dose per fraction to be

utilized over the five fractions. The solution for d in
the above equation is d=2.62 Gy, i.e. 52.62 Gy will
restore the tumour BED10 to that initially prescribed.
Again it should be noted that the required extra BED10
of (50.9)=4.5 Gy10 cannot be added simply pro-rata
across the five 2 Gy fractions. The values of the biological Gy10 and the physical Gy units are dierent and they
cannot be added; to do so would lead to an even higher
fraction dose of 2.9 Gy. This and other subtleties associated with the use BED-equivalents has been discussed in
more depth elsewhere [9,11].
For the normal tissue, the compensated treatment
increases the BED3 to:
BED3 (pre-gap)+BED3 (post-gap)
i.e.

Thus the revised treatment delivers a 6.7% excess in

normal tissue BED3. To evaluate what this compensated
scheme would mean in terms of the equivalent dose
in a schedule delivered with 2 Gy fractions we note
that, by re-arrangement of Eqn (1) and omitting the
repopulation factor:

i.e. the total dose in 2 Gy fractions would be 74.7 Gy.

Thus, the given normal tissue BED3 is approximately
equivalent to 372 Gy fractions.
If the normal tissue dose is considered too high it is
possible to split the dierence, i.e. aim to achieve a
tumour BED10 which is a little less than that prescribed
whilst accepting a small increase in normal tissue BED3.
Such a result may be arrived at by trial and error
processing of dierent values of dose per fraction. For
instance, in the above example an intermediate dose per
fraction of 2.3 Gy would deliver a total tumour BED10
of:
BED10 (pre-gap)+BED10 (post-gap)Tumour
Repopulation Factor
i.e.

BED10 (pre-gap)+BED10 (post-gap)Tumour

Repopulation Factor=Required BED10

The normal tissue BED is:

BED3 (pre-gap)+BED3 (post-gap)

389

i.e.

i.e.

Thus, with 2.3 Gy fractions in the compensation, the

tumour and normal tissue BEDs are respectively 3.5%
lower and 3.1% higher than for the uninterrupted schedule. The eects of higher or lower values of dose per
fraction could be tested, as appropriate, using the same
process. It is stressed that the process of hypofractionating treatment after the gap is not necessarily the best
option: a better result is likely to be obtained if some
extra fractions can be used (via bi-daily fractionation) in
order to further reduce fraction size.

Worked examples for more complex cases

Unscheduled interruptions of longer than 5 days are

generally more dicult to deal with as there is less
chance of completing treatment without incurring a
significant extension of the treatment time. The
following examples highlight such cases.

Example 4. Loss of all of the sixth and seventh weeks

(10 fractions) of a treatment schedule of 70 Gy/35 fractions/
46 days.

As in Example 3, the unscheduled gap runs right up to

the time when treatment should have finished. In this
case however, a very significant part of the treatment has
yet to be delivered. In order to minimize the consequent
extension to treatment time it is inevitable that an
increased dose per fraction will need to be considered if
treatment is to be delivered in once-daily fractions.
We initially attempt to complete treatment in five
fractions delivered during the eighth week, i.e. the
treatment time is extended by 7 days to 53 days. We first
aim to match the prescribed late-normal tissue BED3
(116.7 Gy3), i.e. the dose per fraction to use is d, where d
is solved from:
BED3 (pre-gap)+BED3 (post-gap)=Required BED3.
i.e.

for which d=3.22 Gy.

This same dose per fraction would produce a resultant
tumour BED10 of:
BED10 (pre-gap)+BED10 (post-gap)Tumour
Repopulation Factor

Thus, despite using a relatively large dose per fraction

for the last five fractions, the resultant tumour BED10 is
13.2% less than prescribed. If the weekend prior to the
eighth treatment week is used for treatment then seven
fractions may be delivered, leading to a fractional dose
of 2.57 Gy and a tumour BED10 of 60.1 Gy10. If 11
fractions are distributed over the seven available treatment days (by treating bi-daily on four of them) the
required fractional dose drops to 1.87 Gy, the tumour
BED10 then being 61.9 Gy10. This latter value is still
8.7% short of the prescribed tumour BED10 (67.8 Gy10),
thus some degree of compromise, achieved by increasing
dose per fraction as illustrated in the previous example,
might be considered. In extreme cases thrice-daily
fractionation could be considered, but only after careful consideration of the potential for detriment from
incomplete repair.
If weekend or twice-daily fractionation cannot be
accommodated then it might be considered to treat the
remaining treatment over two full working weeks, i.e.
extend treatment into an eighth and ninth week, making
the overall treatment time 46+14=60 days. For this the
dose per fraction (d) ideally required to maintain the
tumour BED10 is obtained from:
BED10 (pre-gap)+BED10 (post-gap)Tumour
Repopulation Factor
i.e.

for which d=2.85 Gy, leading to an associated BED3 of

138.9 Gy3, which is 19% higher than prescribed. This
result demonstrates the alternative dilemma associated
with further extending the treatment in order to avoid
weekend and twice-daily treatments: the total dose to be
delivered is again increased by the extension into the
ninth week, with a consequently incurred penalty to
BED3.

Example 5. Loss of the final 13 fractions of a treatment

schedule of 70 Gy/35 fractions/46 days.

This represents a very dicult case. As a compromise

between minimizing the extension whilst at the same
time ensuring that a reasonable number of fractions are
used we assume that 10 post-gap fractions will be given,
twice daily from Saturday to Wednesday, extending the

390

treatment to 46+5=51 days. We first consider that the

eect of incomplete repair is negligible, i.e. that Eqn (1)
remains valid. The relevant equation to determine the
dose per fraction (d) to maintain the prescribed normal
tissue BED3 (116.7 Gy3) is:
BED3 (pre-gap)+BED3 (post-gap)=Required BED3
i.e.

For which d=2.41 Gy. The resultant tumour BED10

would then be:
BED10 (pre-gap)+BED10 (post-gap)Tumour
Repopulation Factor
i.e.

To allow for the possibility of incomplete repair in the

critical normal tissue Eqn (2) is used for calculating
the post-gap BED3. Eqn (2) requires prior evaluation of
the h factor, which in turn requires an assumption to be
made about the nature of the repair kinetics. Monoexponential repair half-times for late-normal tissues are
often assumed to be of the order of 1.5 h, but there is
evidence that they may be longer for head and neck
morbidity. Bentzen et al. [31] investigated the repair
half-times of three normal tissue end-points from an
analysis of the CHART head and neck data and found
these to be in the range 3.84.9 h. Taking a mid-range
value of 4.5 h, this corresponds to an exponential repair
rate (
) of 0.15 h
1. (The repair rate is related to repair
half-time via:
=0.693/half-time). If the post-gap daily
fractions are 6 h apart and there is an 18-h overnight
gap, it is easier to calculate h only for the shorter time
interval between any two adjacent fractions. This is
because the incomplete repair after the longer (18 h)
gaps is relatively negligible compared with that following
each 6-h gap. Referring to the Appendix, and using Eqn
(A1) with N=2, =6 h and
=0.15 h
1, h is calculated
to be 0.407. The normal tissue BED3 then becomes from
Eqns (1) and (2) with K=0:
BED3 (pre-gap)+BED3 (post-gap)
i.e.

This is 6.7% higher than the value calculated when

incomplete repair is ignored. If the twice-daily fractions
were to be spaced at 4-h intervals then the h factor is
0.549 and BED3 increases further to 127.4 Gy3, 9.1%
higher than when incomplete repair is ignored. (Because
the h factors are based only on the shorter inter-fraction
intervals the consequent BED3 values are slightly underestimated as there will be an additional amount of
incomplete repair following the longer, overnight, intervals. This overnight contribution will become more
significant as the number of successive days on which
multiple treatment is delivered is increased.)
It has been speculated [33,34] that the sub-lethal
damage repair may not be exponential in form (as
conventionally assumed) but may proceed at a slower
rate than is predicted by a single exponential function. A
reciprocal time model of repair, built on this supposition, has been found to give an excellent fit a to a wide
range of experimental repair data and, unlike models
based on the mono-exponential repair mechanism, helps
explain the cases of radiation myelopathy observed in
the CHART trial [35]. The h factors in the reciprocal
time model may be calculated from Eqn (A2) in the
Appendix and used directly in Eqn (4). Bi-phasic and
other stretched functions have also been developed [36].
Example 6. A nominal 4-week schedule beginning on a
Wednesday is prescribed as 54 Gy/20 fractions/27days. As the
patient is too unwell for the last seven fractions to be treated
on schedule their deferment extends eventual completion of
treatment to 38 days.

This treatment began on a Wednesday and the expected

treatment time (with no treatment at weekends) is 27
days, rather than 25 days if it had began on a Monday.
The prescribed normal tissue BED3 is:

Because the overall time is extended from 27 to 38 days

we assume for calculation purposes that K is zero in the
time up to 28 days and 0.9 Gy day
1 thereafter. The
prescribed tumour BED is therefore:

The interruption extends the overall time to 38 days. If

the seven outstanding fractions were treated at the
original fraction size (2.8 Gy) the late-reaction normal
tissue BED3 would be unaltered. However, the tumour
BED10 will be compromised because the treatment has
extended beyond the 28 days at which time faster
tumour repopulation is assumed to begin.
The tumour BED10 would then be calculated from:
BED10 (pre-gap)+BED10 (post-gap)Tumour
Repopulation Factor


i.e.

391

Conclusion

a reduction of 13.1%.
In short-duration treatments of this type the dose
per fraction is already relatively large and any further
increase (as may be required to strike a balance between
normal tissue and tumour BEDs) should be considered
with caution. As an example, to achieve a tumour
BED10 with an intermediate value 65.0 Gy10 requires a
dose per fraction (d) which is obtained from:
BED10 (pre-gap)+BED10 (post-gap)Tumour
Repopulation Factor=Required BED10
i.e.

i.e. d=3.19 Gy per fraction.

Use of this fraction size for the deferred seven
treatments would increase the normal tissue BED3 to:
BED3 (pre-gap)+BED3 (post-gap)
i.e.

which is still 10% more than that prescribed, even

though the tumour BED10 has been deliberately
compromised.
A final diculty with interrupted schedules which
already employ large fraction sizes is that the scope for
post-gap acceleration using twice-daily treatments is
limited on account of the large total daily doses which
would result. In this particular example, bi-daily fractionation would deliver a total daily dose of 22.7 Gy,
corresponding to a BED3 delivery rate of 10.3 Gy3 per
day. This is over 50% higher than the BED3 delivery rate
(6.8 Gy3 per day) associated with the thrice-daily fractionation of CHART, which itself is very similar to
22 Gy daily fractionation (6.7 Gy3 per day). These
figures take no account of incomplete repair, which
would increment the BED3 associated with 22.7 Gy
even further. Even with well-spaced fractions some
caution would be required when contemplating daily
biological doses of this magnitude and the possibility of
treating bi-daily with a significantly reduced fractional
dose should be explored.

This paper sets out the practical considerations involved

in calculating compensations once a treatment interruption has occurred, but it is stressed that it is good
practice to aim to avoid interruptions wherever possible.
A wealth of radiobiological and clinical evidence confirms the seriousness of unnecessary treatment prolongation and unscheduled gaps, particularly those near the
end of treatment, are especially problematic. Although
it is impossible to pre-empt some interruptions, e.g.
those resulting from machine breakdowns or bad clinical
response to treatment, those resulting form Public
Holidays and Statutory Days can be pre-planned. It is
not sucient to devise last-minute compensations for
such interruptions and departmental policies should
prohibit such practice. Rather, patients should be categorized in advance according to RCR guidelines and
their treatment schedules reviewed such that the predictable treatment gap is properly compensated by, if
necessary, revision to the whole schedule.
Extremely busy departments are clearly more likely to
be troubled by unscheduled gaps and will also find it less
easy to timetable an eective compensation when such
gaps do occur. The permanent presence of spare treatment capacity (e.g. by provision of enough treatment
machines within each department to allow the deliberate
under-usage of at least one of them) is an obvious
method for anticipating and coping with unscheduled
interruptions. Such a policy has resource implications,
but these need to be set against the fact that with
treatment interruptions, as in several other aspects of
radiotherapy treatment delivery, there exists a likely
inverse relationship between pressurized working
conditions and treatment quality.
Radiobiological methods of compensation, when they
have to be utilized, are better than none at all, but it is
nevertheless essential to be aware of the potential downside of invoking calculations which are themselves oversimplifications and which rely on parameter valueswhich
sometimes are not accurately known. Three particular
aspects of the calculation methods set out above should
be borne in mind.
Firstly (and as has been assumed in the calculations
here), although there is strong evidence for accelerated
repopulation following an initial period of slower
repopulation, it is not clear whether or not the K factor
varies during treatment. There is the possibility that
changing the dose per fraction, or the dose delivery rate,
(as occurs in devising some gap compensations) may
itself alter K or the lag period.
Secondly, the sub-lethal damage repair mechanisms
are not fully understood and there is a strong likelihood that repair is slow in some normal tissues.
Hyperfractionated compensation schemes, involving
the delivery of many closely-spaced fractions after the
unscheduled gap, carry the risk of significantly increasing normal tissue morbidity, even in cases where the
consequences of incomplete repair potentially have been

392

allowed for by the inclusion of a seemingly appropriate

h factor in the calculations.
Finally, it should be remembered that BED calculations take no account of the physiological factors which
may influence radiation response throughout an irradiated volume. The response is essentially governed by the
integrated eect of the radiation, which in turn is related
to the degree of dose uniformity and to the functional
complexity of the tissue or organ. Where there are
unavoidable or deliberate dose gradients the response to
the gap compensation may dier throughout the treatment volume. In such cases it is prudent to repeat the
biological assessments at the normal tissue hot spots
[37]. Where tumour shrinkage is known to have
occurred the clinician may feel that the use of reduced
field size during gap compensation may partially oset
some of the potential problems. The possibilities of this
and other safeguards (increased shielding, revised treatment plan, etc.) should always be considered alongside
the radiobiological aspects whenever the normal tissue
BED is likely to increase as a result of treatment
prolongation.
Ongoing elucidation of radiobiological processes and
parameters will inevitably mean that there will be a
requirement to review and, in some cases, modify, the
recommendations set out in this article. It is recommended that each radiotherapy centre identifies an individual (local or remote) who is in a position to keep
abreast of unfolding developments and who can be the
prime source of advice to that centre on how to devise
treatment compensations.
Acknowledgements. The authors wish to express their appreciation to Drs Colligan (Inverness), Fowler (Wisconsin), Henk (London),
Morgan (Nottingham), Roberts (Manchester), Slevin (Manchester)
and Spittle (London), all of whom read the draft manuscript and made
a number of useful and constructive comments.

10
11

12

13

14
15

16

17

18

19

20

REFERENCES
1 Hendry JH, Bentzen SM, Dale RG, et al. A modelled comparison
of the eect of using dierent ways to compensate for missed
treatment days in radiotherapy. Clin Oncol 1996;8:297307.
2 Slevin NJ, Hendry JH, Roberts SA, Agren-Cronqvist A. The
eects of increasing the treatment time beyond three weeks on
the control of T2 and T3 laryngeal cancer using radiotherapy.
Radiother Oncol 1992;24:215220.
3 Fowler JF, Chappell R. Non small cell lung tumours repopulate
rapidly during radiation therapy. (Letter) . Int J Radiat Oncol Biol
Phys 2000;46:516517.
4 Fowler JF. Radiobiological modelling to compensate for
unplanned gaps in radiation treatment. International Congress
of Radiation Research, Dublin, July, 1999. In Radiation Research,
Vol 2 Proceedings. Moriarty M, Mothersill C, Seymour C,
Eddington M, Ward JF, Fry RJM, eds. Oak Brook, IL, U.S.A.:
Int. Assoc. Rad. Res., pp. 653659, 2000.
5 RCR Ref No BFCO (96) 4. Guidelines for the management of
the unscheduled interruption or prolongation of a radical course
of radiotherapy. ISBN 1 872599 26 5. 1996; Royal College of
Radiologists.
6 Sinclair JA, Oates J, Dale RG. BED-time charts and their application to the problems of interruptions in external beam

21

22

23

24

25

26

radiotherapy treatments. Int J Radiat Onc Biol Physics 1999;

44:381389.
Wheldon TE, Barrett A. Radiobiological rationale for compensation for gaps in radiotherapy regimes by post-gap acceleration of
fractionation. Br J Radiol 1990;63:114119.
Dale RG, Sinclair JA. A proposed Figure of Merit for the
radiobiological assessment of unscheduled treatment interruptions.
Br J Radiol 1994;67:10011007.
Dale RG, Jones B, Sinclair JA. Dose-equivalents of tumour
repopulation during radiotherapy: the potential for confusion. Br J
Radiol 2000;73:892894.
Doerr W, Hendry JH. Consequential late eects in normal tissues.
Radiother Oncol 2001;61:223231.
Jones B, Dale RG, Deehan C, et al. The role of biologically
eective dose (BED) in clinical oncology. Clin Oncol 2001;13:
7181.
Withers HR, Taylor JMG, Maciejewski B. The hazard of accelerated tumour clonogen repopulation during radiotherapy. Acta
Oncol 1988;27:131146.
Fowler JF, Harari PM. Confirmation of improved loco-regional
control with altered fractionation in head and neck cancer. Int J
Radiat Oncol Biol Phys 2000;48:36.
Steel GG. (ed.) Basic clinical radiobiology. London: Arnold, 1997.
Roberts SA, Hendry JH. Time factors in larynx tumour radiotherapy: Lag times and intertumour heterogeneity in clinical datasets from four centres. Int J Radiat Oncol Biol Phys 1999;45:
12471257.
Fu KK, Pajak TF, Trotti A, et al. A Radiation Therapy Oncology
Group (RTOG) Phase III randomised study to compare hyperfractionation and two variants of accelerated fractionation to standard
fractionation radiotherapy for head and neck squamous cell carcinomas: first report of RTOG 9003. Int J Radiat Oncol Biol Phys
2000;48:716.
Withers HR, Peters LJ. Transmutability of dose and time.
Comments on the 1st report of RTOG 9003 (KK Fu et al) . Int J
Radiat Oncol Biol Phys 2000;48:12.
Roberts SA, Hendry JA. The delay before onset of accelerated
tumour cell repopulation during radiotherapy: a direct maximumlikelihood analysis of a collection of worldwide tumour-control
datasets. Radiother Oncol 1993;29:6974.
Allal AS, de Pree C, Dulguerov P, Bierei S, Maire D, Kurtz JM.
Avoidance of treatment interruption: an unrecognised benefit
of accelerated radiotherapy in oropharyngeal carcinomas? Int J
Radiat Oncol Biol Phys 1999;45:4145.
Robertson C, Robertson AG, Hendry JH, et al. Similar decreases
in local control are calculated for treatment protraction and for
interruptions in the radiotherapy of carcinoma of the larynx in four
centers. Int J Radiat Oncol Biol Phys 1998;40:319329.
Robertson AG, Robertson C, Perone C, et al. Eect of gap
length and position on results of treatment of cancer of the
larynx in Scotland by radiotherapy: a linear quadratic analysis.
Radiother.Oncol 1998;48:165173.
Baumann M, Petersen C, Wolf J, Schreiber A, Zips D. No evidence
for a dierent magnitude of the time factor for continuously
fractionated irradiation and protocols including gaps in two
human squamous cell carcinoma in nude mice. Radiother Oncol
2001;59:187194.
Zieron JO, Omniczynski M, Raabe A, Beck-Bornholdt HP. Impact
of the position of a gap on the response of the RIH-tumour of the
rat on fractionated split-course radiotherapy. Strahlenther Onkol
2001;177:592596.
Sanguineti G, Benasso M, Corvo R, et al. Lack of a time factor in
alternating chemotherapy for advanced head and neck squamous
cell carcinoma. Tumori 2001;87:1013.
Weber DC, Kurtz JM, Allal AS. The impact of gap duration on
local control in anal canal carcinoma treated by split-course
radiotherapy and concomitant chemotherapy. Int J Radiat Oncol
Biol Phys 2001;50:675680.
Bentzen SM, Thames HD. Overall treatment time and tumour
control dose for head and neck tumours: the dog leg revisited.
Radiother Oncol 1992;25:143144.


27 Mehta M, Scrimger R, Mackie R, et al. A new approach to dose
escalation in non small cell lung cancer. Int J Radiat Oncol Biol
Phys 2001;49:2333.
28 Haustermans K, Fowler J, Geboes K, et al. Relationship between
potential doubling time (Tpot), labelling index and duration of
DNA synthesis in 60 oesophageal and 35 breast tumours: is it
worthwhile to measure Tpot? Radiother Oncol 1998;46:157167.
29 Trott KR, Kummermehr J. What is known about tumour proliferation rates to choose between accelerated fractionation or
hyperfractionation? Radiother Oncol 1985;3:19.
30 King CR. What is the Tpot for prostate cancer? Radiobiological
implications of the equivalent outcome with 125I or 103Pd. Int J
Radiat Oncol Biol Phys 2000;47:11651167.
31 Bentzen SM, Saunders MI, Dische S. Repair halftimes estimated
from observations of treatment-related morbidity after CHART or
conventional radiotherapy in head and neck cancer. Radiother
Oncol 1999;53:219226.
32 Thames HD, Hendry JH. Fractionation in radiotherapy. London:
Taylor and Francis, 1987.
33 Fowler JF. Is repair of DNA strand break damage from
ionizing radiation second order rather than first order? A simpler
explanation of apparently multiexponential repair. Radiat Res
1999;153:124136.
34 Dale RG, Fowler JF, Jones B. A new incomplete-repair model
based on a reciprocal-time pattern of sub-lethal damage repair.
Acta Oncol 1999;7:919929.
35 Dische S, Saunders MI. Continuous, hyperfractionated, accelerated radiotherapy (CHART): an interim report upon late
morbidity. Radiother Oncol 1989;16:6572.
36 Millar WT, Van den Aardweg GJ, Hopewell JW, Canney PA.
Repair kinetics in pig epidermis: an analysis based on two separate
rates of repair. Int J Radiat Biol 1996;69:123140.
37 Jones B, Dale RG, Khaksar SJ. Biological equivalent dose assessment of the consequences of hypofractionated radiotherapy. Int J
Radiat Oncol Biol Phys 2000;47:13791384.

APPENDIX
Calculation of h values for closely-spaced fractions

For mono-exponential recovery kinetics h is calculated

from:

393

In Eqn (A1), N is the number of fractions in a group of

closely spaced fractions and =exp(

x), where
is the
exponential repair constant of sub-lethal damage and x
is the average time interval (h) between these closelyspaced fractions. For late-responding normal tissues
is
often assumed to have a generic value of 0.5 h
1; for
prediction of radiation myelopathy subsequent to head
and neck radiotherapy a smaller value of around
0.22 h
1 is more appropriate [31,32].
For reciprocal-time repair kinetics h is calculated
from:

In this model z is the reciprocal-time repair constant of

sub-lethal damage [34]. The myelopathies observed in
the CHART Trial [35] are consistent with a z value of
0.36 h
1 [34].
Provided there are only a few consecutive days on
which multiple daily fractions are delivered the calculation of h using either Eqn (A1) or Eqn (A2) may be
simplified by taking into account only the shortest
inter-fraction intervals. For example, if twice-daily fractionation is being used with an interval of 6 h between
fractions, followed by an longer overnight interval of
18 h, the values used to calculate h are N=2 and x=6.
Allowance for the additional incomplete repair following the longer overnight intervals may become necessary
if there are many successive days on which twice-daily
fractions is delivered. Reference [34] provides more
details on this.