Breast Cancer Research and Treatment60: 201–209, 2000.© 2000Kluwer Academic Publishers. Printed in the Netherlands.

Report

The timing of treatment in breast cancer: gaps and delays in treatmentcan be harmful

Ann JohnsonBreast Study Centre, Mount Vernon Hospital, Northwood, UK

Key words:breast cancer, growth rates, histological grade, primary medical treatment, radiotherapy fractionation,‘split-course’ radiotherapy

Summary

‘Timing’ of treatment in breast cancer may refer to intervals within a single management or between differentmanagements. Rates of shrinkage of breast cancers in response to treatment are related to histological grade andmay be used as surrogates for growth rates. Histological grade should predict appropriate timing of treatment.Four cases of locally advanced breast cancer that illustrate a number of different types of interval are presented.Two tumours of differing histological grade (II and III) had been managed by historical ‘split-course’ radiotherapyand two similar grade III tumours were managed by primary medical treatment, followed at different intervals byradiotherapy. In the grade III tumours different radiotherapy fractionation régimes and effects of varying intervalsbetween mangements are compared. The theoretical advantage of shrinkage (leading to reoxygenation) during thegap in ‘split-course’ radiotherapy is realized only in relatively slowly growing and shrinking tumours. Grade IIItumours grow rapidly. They have the potential to shrink rapidly in response to appropriate treatment, namely in-tensive chemotherapy or radiotherapy but not hormones. Inadequate treatment leads to growth in intervals betweenindividual doses, whether of drugs or radiation, and to failure of local control. The advantage of surgery or primarymedical treatment will be lost if the interval between managements is too long in relation to the volume doublingtime. Histological grade is a good guide of this parameter; the grade III tumours are particularly vulnerable to gapsin treatment.

Introduction

In the absence of a conceptual image of individualwhole tumour behaviour, the timing of multidiscip-linary treatments in breast cancer is haphazard. Treat-ments are prescribed with due consideration of theirtoxicity and of the convenience of doctor and patient.Surgeons like to see a healed wound before referringtheir patients on for managements known to inter-fere with healing. In single managements, non-specifictreatments like chemotherapy are usually prescribed atintervals that allow recovery of normal tissues such asthe bone marrow.

Radiotherapists have traditionally allowed breaksat the weekend and on public holidays without ad-justment of the dose. Requests to postpone treatmentsfor social reasons are seldom refused. The chief diffi-

culty in deciding upon timing is that evidence aboutwhat is actually happening to the tumour is scanty.For instance, allocation to post-operative radiation isbased upon statistical probabilities derived from over-all recurrence rates in its absence [1, 2]. Precisemeasurement of residual tumour is lacking. There willbe a range from tumour-free to barely sub-clinicalmasses for which the same post-operative radiother-apy may have been prescribed. A cost-effectivenessequation that includes tumour control and long-termtoxicity is hard to derive in the individual case [3]. Ifthere is a waiting list for treatment, does this matterand to which patients?

Much can be learned from the management of loc-ally advanced tumours. When the only palliation forinoperable tumours was radiotherapy, it was recog-nized that the inevitable hypoxia within such tumours

202 Ann Johnson

increased the dose of radiation required to ‘cure’ andthat the larger the tumour, the larger this dose willbecome, exceeding normal tissue tolerance [4, 5].Fractionation of radiotherapy appears to overcomemuch hypoxia. A modification was the suggestionthat a gap should be left in the course of treatmentduring which shrinkage and reoxygenation might oc-cur [6]; it is sine qua nonthat the tumour shouldshrink. Latterly, it has been recognized that tumoursdo not stop growing on the first day of treatment andthat uncompensated gaps in radiotherapy may lead toloss of local control [7, 8]; techniques of accelerated,hyperfractionated radiotherapy have been introducedin an attempt to overcome early repopulation duringtreatment [9]. Again, if repopulation does not occur,the altered fractionation may be a disadvantage. Itis, therefore, necessary to be able to predict tumourdynamics before experimenting with timing. Measure-ment of the rate of shrinkage of unselected primarybreast cancers in response to primary medical treat-ment [10] has demonstrated that the intrinsic rate foran individual tumour may be predicted from its his-tological grade. Grade III tumours grow and shrinkquickly, grade I tumours grow and shrink slowly withgrade II in the intermediate position. This conclusionproved elusive because there was also a clear cor-relation between rates of shrinkage and the type oftreatment by which it had been generated [11]. Whilesome treatments are simply less effective than othersin generating the full range of rates of shrinkage, someare specific for a tumour type. The clearest illustrationof the latter is in relation to hormone dependency –well-differentiated, hormone-dependent tumours werefound to shrink relatively slowly. Shrinkage rates havebeen successfully used as surrogates for growth ratesin a model to predict the incidence of interval cancers[12]. Histological grade should therefore be a valuableindex of the probable growth rate.

Examples of planned ‘split-course’ radiotherapy intwo different tumours and of primary medical treat-ment followed radiotherapy in two similar tumourswhere different variations of the timing of treatmentoccurred, will be examined.

Patients

All four patients had inoperable primary breast can-cers. The main presenting findings are shown inTable 1. Cases 1 and 2 were selected for the ‘splitcourse’ radiotherapy by virtue of the locally advanced

Table 1. Clinical presenting features

Case Age Histology Grade Diameter

1 74 Ductal II 68 mm

2 53 Ductal III 78 mm

3 59 Ductal III 31 mm

4 39 Medullary III 35 mm

state of the tumours. The biopsy in case 2 showedgross necrosis (Figure 1). Case 3 was a thin lady withvery small breasts whose lower, inner quadrant tumourwas already fixed deeply with reddening of the overly-ing skin. In case 4, who was the only premenopausalpatient, the tumour was similarly situated and fixed.This patient was short in stature and obese.

Treatments

Radiotherapy

The details of the radiotherapy treatments are givenin Table 2. The ‘split course’ radiotherapy for cases 1and 2 and the irradiation of case 3 were administered,according to normal practice in 1982, by colleaguesin the Regional Radiotherapy Centre; there was no al-teration of dose to compensate for gaps in treatment,however caused. Radiotherapy for case 4 was planned‘in house’; the dose was calculated to give the equiv-alent of 60 Gy in 6 weeks using Orton’s modificationof the Ellis formula [13]. The biologically effectivedoses (BED) with correction for overall time and pro-liferation have now been calculated using the linearquadratic formula [14]

BED= nd(

1+ d

α/β

)− log 2

αxT

Teff,

wheren = no. of fractions,d =dose per fraction,T =overall time andTeff = effective doubling time [15,16]. Theα/β ratio was assumed to be 10 for tumourand 3 for late reacting tissues.α was taken as 0.3 andTeff as 20 days for case 1 and 6 days for cases 2, 3and 4. The normal tissue doses refer to that part of thetreatment that was directed at the whole chest wall.The total doses within restricted fields applied as ‘top-up’ to the tumour sites in cases 3 and 4 are shown inparentheses.

Systemic therapy

Primary medical treatment took place in the BreastStudy Centre between 1983 and 1985. Each patient

Timing of treatment 203

Figure 1. Drill biopsy of case 2 before treatment, x100, stained with haematoxylin and eosin. Extensive necrosis at a distance of about 200microns from the capillary contrasts with active proliferation nearby.

received a full explanation of the protocol before giv-ing her consent to take part in the study. Since theoriginal hypothesis was that tumours would shrink attheir own intrinsic rates, that would be dictated bytheir structure and growth characteristics irrespectiveof the form of treatment, we sought the least toxicchemotherapy that would achieve this rate for an in-dividual tumour. It was proposed that the reduction involume that might be achieved by a combination ofthree drugs followed by a three week interval for bonemarrow recovery (the then current protocol) mightequally well result from the weekly injection of onedrug, accompanied by lower toxicity. A second vari-ation from standard management was that the problemof drug resistance might be alleviated by the sequen-tial administration of single agents in short courses. Inits final form, a ‘course’ was defined as four weeklyintravenous injections of a single chemotherapeuticagent: fluorouracil 600 mg/sq.M (F), cyclophospham-ide 300 mg/sq.M (C), vincristine 1 mg (Vc), metho-trexate 15 mg/sq.M (M) or doxorubicin 15 mg/sq.M

(A) (replaced by mitozantrone 3 mg/sq.M after it hadbeen introduced). The first course was followed by atrial of tamoxifen 40 mg daily (Tf). If the latter provedeffective, it was continued and a gap of up to sixweeks permitted between courses of chemotherapy.Failure to continue shrinking, or regrowth, shortenedthe interval between courses or resulted in a change ofmanagement. Where the single agent/hormone regimeresulted in continuing growth, recourse was made tocombination therapy at weekly intervals: fluorouracil1000 mg, cyclophosphamide 500 mg and vincristine1 mg (FCVc). This latter method had been used effect-ively in locally advanced breast cancers at a time whenthere was a waiting list of six weeks for radiotherapy[11] – it had not been designed as a long-term manage-ment because of its toxicity. During the limited periodof its application, arrangements would be set in trainfor surgery and /or radiotherapy to follow. There wasnow no question of a planned gap between treatmentssince the situation was one of the observed failure ofcontrol.

204 Ann Johnson

Table 2. Radiation treatments

Case 1 Case 2 Case 3 Case 4

1st radiation 5 MeV Co60 Co60 5 MeV

Dose per fraction Gy 2.4 2 2.67 (×14) 2.67

plus 3 (×2)

No. of fractions 13 13 16 20

Days 17 17 24 29

Gap 21 17 0 21

2nd radiation 5 MeV Co60 ∗10 MeVe− ∗Co60

Dose per fraction Gy 2.4 2 (×11) plus 2.86 2.5

3 (×4)

No. of fractions 12 15 7 6

Days 19 19 12 8

Total fractions 25 28 23 26

Overall time days 57 52 36 58

Total tumour dose Gy 60 60 63.4 68.4

BED10 74 74 81 87

BED10 corrected 68 53 67 64

BED3 108 104 83 (122) 101 (129)

BED: Biologically Effective Dose, with subscript identifyingα/β ratio.∗Reduced field sizes localized to tumour. Figures in parentheses show BED3 formaxima in reduced fields.

Measurement

Progress of treatment was monitored by calliper meas-urements of tumour diameters. The tumour volumeswere plotted on a log10 scale, promising data sets weresubjected to an analysis of variance and regressionlines fitted by a method of weighted least squares tosignificant data sets only. No line was fitted to less thanfour points. Apparently similar lines were tested todiscover whether parallel lines were an improvementover separate single lines. From the lines the volumehalving time (vht) may be calculated [17].

Results

The measurement data are shown in Table 3 and illus-trated in Figures 2 to 5. Where a figure has two graphsthe upper one shows the whole measured time courseand the lower is an enlargement of a significant gaparea. All the regression lines are highly significant.The wide confidence intervals in case 2 are the resultof the small number of data points.

Case 1 (Figure 2)

Shrinkage started at the beginning of treatment andcontinued through the gap, so that one line is fit-

Table 3. Measurement data

Case 1 Case 2 Case 3 Case 4

Treatment RTa RT F//CVc//RT F//IFCVc//RT

Slope −0.0085 −0.0092 −0.029 −0.051

F 220.4 19.7 65.1 72.2

p <0.0001 <0.0001 <0.0001 <0.0001

SEb 0.0006 0.0021 0.0025 0.0037

vht days 33 35 10 6

95% conf. 31–41 22–64 8.8–13 5–7

aRT: Radiotherapy.bSE: Standard error of estimate.

ted through to the middle of the second part of theradiation. At this point, skin reaction precluded meas-urement. When measurement was again possible, asecond line is found to be parallel with the first butstarting from the same volume that had been reachedat the end of radiation. Local control was not achieved;death was due to pulmonary metastases at 21

2 yearsfrom diagnosis.

Case 2 (Figure 3)

No significant shrinkage occurred until the secondweek of the second course of irradiation. The dose

Timing of treatment 205

Figure 2. Time course of case 1. RTl and RT2 show ‘split course’radiotherapy; F, fluorouracil; mg, medroxyprogesterone.

per fraction was raised from 2 Gy to 3 Gy at fraction24 but the patient had by then developed a severeskin reaction so that there are no measurements atthis point of interest. Local control was not achieved.Death from cerebral metastases occurred 10 monthsafter diagnosis.

Case 3 (Figure 4)

After satisfactory shrinkage due to fluorouracil, thispatient’s tumour enlarged rapidly on tamoxifen. Theoverlying skin, that had become normal in appear-ance during the chemotherapy, became inflamed andtender. Cyclophosphamide proved effective and thetumour disappeared from view. Regrowth at day 250was stopped by repetition of fluorouracil but othersingle agents did not work. Treatment was changedto weekly FCVc with shrinkage at the same rate as theoriginal response to fluorouracil alone. Arrangementswere made to start radiotherapy immediately. In theevent, an unplanned gap of 13 days without treatmentand 16 days without measurement occurred. When

Figure 3. Time course of case 2. RTl and RT2 show ‘split course’radiotherapy; F, fluorouracil; C, cyclophosphamide.

measurement was restarted the tumour had enlargedby nearly 3 volume doublings to a greater size thanthe original pre-treatment volume. The radiotherapythen produced shrinkage at the same rate as that dueto chemotherapy but local control was not achieved.Recurrence was managed by salvage mastectomy andfurther radiotherapy with hyperthermia to an axil-lary recurrence in order to achieve long term control.This patient remains alive and well at 14 years fromdiagnosis.

Case 4 (Figure 5)

Overall, chemotherapy and tamoxifen allowed contin-ued growth to which a regression line was originallyfitted [10]. (It is also possible to fit a separate, steeperline to the growth during tamoxifen administration).Treatment was changed to weekly FCVc with im-mediate effect. However, vincristine neuropathy wasvery troublesome and the omission of this one drugimpaired the steep shrinkage.

206 Ann Johnson

Figure 4. Time course of case 3. F, fluorouracil; C, cyclophosph-amide; Vc, vincristine; M, rnethotrexate; A, doxorubicin; FCVc,cornbination of drugs; Tf, tamoxifen; RT, radiotherapy.

Radiotherapy followed one week after the lastchemotherapy, resulting in rapid clinical disappear-ance of the tumour. Because of her size and shape

Figure 5. Time course of case 4. M, methotrexate; A, doxorubicin;F, fluorouracil; C, cyclophoshamide; Vc, vincristine; FCVc, com-bination of drugs; RT1, radiotherapy to whole breast; RT2, boost totumour site.

the treatment design necessarily delivered more thantissue tolerance doses to part of the lateral chest walland a tolerance dose to part of the liver. Predictably,she has developed fibrosis in the lateral chest wall andareas of fatty infiltration in the liver. Long term localcontrol has been achieved with little disability. She isalso alive and well at 12 years from diagnosis.

Discussion

Two different aspects of timing are illustrated by thesecases. First, we may look at gaps in radiotherapyschedules, whether accidental or planned. Secondly,we have examples of the effect of variations in spacingwhen the treatment is changed.

The key to the interpretation of the results lies inthe two ‘split-course’ cases. Because of the large sizeof the tumours, we are able to see the whole courseof the radiation response, although it was unfortu-nate that skin reactions resulted in some loss of data.The pattern of shrinkage in case 1, with a vht of 33days, appears to be a little slow for a grade II tumour(Table 4). However, we have demonstrated that verylarge tumours do shrink more slowly, probably dueto circulatory problems and to the accumulation of alarge amount of non-cellular material [11]. TheTeff of20 days was chosen because there was no regrowth inthe gap. It would appear that reported potential doub-ling times (Tpot) in breast cancer [18] are weighted bythe rapidly growing grade III tumours that are easierto disaggregate for cell flow cytometry. Also, rapiddivision will not be followed by equally rapid growthwhen there is much loss from the clonogenic cell com-partment through differentiation. In case 2, the failureto shrink until near the end of treatment is a phe-nomenon that has also been observed: analysis of datafrom 68 patients where radiation was the first treat-ment showed that the greater the cumulative radiationeffect (CRE) before the start of shrinkage, the morelikely it was that the ensuing slope would be steep

Table 4. Volume halving times according tohistological grade [10]

Grade No. of tumours Mean vht days

I 34 33 (25–44)

II 104 23 (20–27)

III 58 14 (12–17)

Figures obtained from 196 tumours, figures inparentheses are 95% confidence limits.

Timing of treatment 207

[11] – that is, it would probably be grade III histology[10]. The vht of 35 days is clearly an overestimate.The failure to shrink resulted in a decision to increasethe dose per fraction but too late to obtain measure-ments. Close inspection of the measurements that weremade on five consecutive weekdays reveals apparentlyrapid shrinkage, followed by an upturn at the week-end break (Figure 3, lower graph, 2nd radiation). Thispattern has previously been reported in a grade IIItumour where more rapid shrinkage due to weeklycombination chemotherapy showed that the radiation-induced shrinkage (2 Gy per fraction with gaps intreatment) was not reaching the maximum possiblerate [11].

Cases 3 and 4 happened to receive higher doses perfraction than case 2 and the resulting rates of shrink-age were appropriate to their histological grade III. Interms of BED their irradiation was similar. The as-sumption of aTeff of 6 days for the grade III tumoursis supported in two ways. First, the growth during theunintended gap in the treatment of case 3 resulted ina doubling of the tumour volume every 5 days; andsecondly, the plateau during single agent chemother-apy in case 4 can be interpreted as removal of half thevolume and its replacement from the surviving clono-genic cells with a volume doubling time of 7 days.(The histology of case 2 conforms to Fowler’s pictorialmodel that was based uponTpot values of 4–6 daysin squamous carcinomas of the head and neck [15].)The failure of local control in case 3 may thereforebe attributed the large size, nearly 35 mm diameter, atthe start of radiotherapy. It was still 15 mm diameterwhen all the radiation had been delivered. The largerthe tumour, the greater the dose that will be requiredto eliminate all clonogenic cells and the greater is thelikelihood that hypoxic cells were sequestered in theresidue. In case 4 the radiotherapy starting diameterwas only 20 mm and the tumour had disappeared fromclinical view afier six treatments. At that rate, the tu-mour diameter might be about 7 mm diameter at theend of the first radiotherapy. It is possible that steril-ization of clonogenic cells had already been achievedat this point (BED10 = 56 before the gap). The BED3was high in all cases; in cases 1 and 2 the full coursewas delivered to the whole breast but in cases 3 and 4it was possible to restrict the ‘top-up’ dose to a lim-ited area. (BED3 shown in parentheses in Table 2).The severely damaged tissue was removed surgicallyin case 3 but case 4 demonstrates the familiar linearfibrosis at the site of maximum build up of tangen-tial fields. These two tumours appear to have been

suitable for accelerated hyperfractionation, that mighthave spared the normal tissue damage.

The effect of gaps between managements is seenin case 3. Despite the failure of tamoxifen, long in-tervals were left between courses of single agents.The resulting regrowth was then not halted by closerspacing. This suggests that regrowth on single agentchemotherapy was due to a combination of over-widespacing and low doses; it was not due to drug res-istance since the same drugs in combination, at thesame individual doses, were subsequently effective. Ifthe large starting size at the beginning of radiotherapywas the mechanism of failure of local control then thecause was the inadvertent gap after chemotherapy. Inthis short space of time any advantage that might havebeen conferred by chemotherapy was negated. In case4, once reduction had been started by suitably intens-ive chemotherapy, a gap before radiotherapy wouldhave been as detrimental as in case 3. We can say thiswith confidence because the loss of one drug from thecombination halted the shrinkage. The inappropriateuse of hormones also constitutes a gap in effectivetreatment.

It would appear that only case 4 achieved the max-imum possible rate of shrinkage. It is likely that in thegrade III tumours the clonogenic fraction was large,dominating patterns of growth and shrinkage, so thatTeff approximates to the volume doubling time. Inthe well-differentiated grade II tumour the pattern ofshrinkage would be dictated by the rate of death andremoval of differentiated cells and stromal compon-ents, until such time as the regrowth from residualclonogenic cells exceeded the losses.

The normal distribution of rates of shrinkage froman unselected series of 98 new primary breast can-cers has been used as a surrogate for growth ratesin a model that successfully predicted interval cancerincidence [12]. The mean volume halving/surrogatedoubling time in that series was 1 month. This is muchmore rapid than the average 90 day doubling timethat is found in the literature [19]. Unfortunately, alimited selection of tumours lends itself to measure-ment of growth so that case reports are dominated byslowly growing tumours. We have estimated that up to40% of tumours may have volume doubling times thatare shorter than the 3 week intervals between stand-ard chemotherapy treatments; for 2 weeks this figurewill be 21% and for 1 week 5% [10]. Tumours likethe three grade III tumours that have been presentedmay also grow significantly in gaps in radiotherapydue to public holidays or equipment failure, gaps to

208 Ann Johnson

allow resolution of reactions and ‘split-course’ treat-ments [8, 17]. Most breast cancers are now relativelysmall at presentation; many will have been removedas an abnormality found on screening. The absence ofvisible tumour after local excision generates a senseof false security. The probability of local control maybe reduced by delays between surgery and radiother-apy that are imposed by waiting lists for appointmentsand treatment; rapidly growing tumours will be mostvulnerable.

The broad spectrum of growth rates introducesa problem in relation to randomized controlled clin-ical trials. Without selection for treatment accordingto probable growth rate, the likelihood of discover-ing the best treatments for individual tumours willbe very much reduced [20, 21]. Intensive chemother-apy may be ‘overkill’ for tumours that shrink slowlyand different radiotherapy fractionation may be neces-sary for rapidly growing tumours. A single protocolis most unlikely to be correct across the wide rangeof tumour behaviour. Investigation of new radiother-apy fractionation schedules should be focussed byselection according to likely tumour kinetics [14].Potential doubling times measured by cell flow cyto-metry do not reflect clinical volume doubling timesfor most tumours; however, immunohistochemistry ofbromodeoxyuridine-labelled cells has revealed spatialdistribution patterns that may add to the predictivevalue of flow cytometry [22, 23].

Conclusions

The timing of treatment is important. The range ofrates of shrinkage of breast cancers in response totreatment, that are closely related to their growth rates,demands that differing protocols be applied to differ-ent tumours. The cases that have been presented in thispaper illustrate the importance of histological grade asa guide. The theoretical advantage of re-oxygenationduring gaps in treatment can be realized in relativelyslowly growing and shrinking tumours of grades I andII. Tumours whose intrinsic shrinkage rates are slowcan be managed by less intensive régimes, with fewerside effects. Rapidly growing grade III tumours havebeen shown to be unsuitable for ‘split-course’ radio-therapy schedules, schedules where no allowance ismade for proliferation during gaps and probably evenfor the time-honoured 2 Gy daily radiation doses perfraction. These strictures apply equally to systemictherapy where intensive high dose chemotherapy is

needed for the rapidly growing tumours. Gaps withinor between treatments may lead to loss of local con-trol. Where there are waiting lists, priority shouldbe given to such cases. Better still, in a dedicatedbreast clinic the whole management schedule may bemapped out before treatment, so that each phase flowssmoothly in to the next.

Acknowledgements

Figure 5 is reproduced (slightly altered) by the kindpermission of the Editor ofThe Breast.

The measurements in all cases, the primary med-ical treatment of case 3 and the whole managementof case 4 took place in the Breast Study Centre,Mount Vernon Hospital, between 1982 and 1985, un-der the leadership of the late Dr. R.H. Thomlinson. Weare grateful to Dr. Paul Strickland, former Chairmanof the Regional Radiotherapy Centre and Dr. Eliza-beth Grosch, present Clinical Director of the Centrefor Cancer Treatment for referring their patients formeasurement and to Dr. Jane Maher, Consultant Clin-ical Oncologist, Centre for Cancer Treatment andDr. Clare Vernon, Consultant Radiotherapist and On-cologist, Hammersmith Hospital for the combinedmanagement of the recurrence in case 4. The useof computer facilities of the Gray Laboratory CancerResearch Trust is gratefully acknowledged.

The Breast Study Centre was originally funded bythe Medical Research Council, of whose external sci-entific staff Dr. Thomlinson was a member. Since 1985the work has been supported solely by charitable dona-tions through the Mount Vernon Breast Cancer ClinicAppeal, Chairman the late Mrs. Jan Birdsall, Treasurerthe late Mr. John Smith.

References

1. Rosen PP, Fracchia AA, Urban JA, Schottenfeld D, Rob-bins GF: ‘Residual’ mammary carcinoma following simulatedpartial mastectomy. Cancer 35: 739–747, 1975

2. Morgan DAL, Galea MH, Berridge J, Blamey RW, Elston CW,Ellis IO: Post-mastectomy radiotherapy in patients at high riskof locoregional recurrence: a randomized trial. The Breast 1:151, 1992

3. Dische S, Joslin CAF, Miller S, Bell NL, Holmes JC: Thebreast radiation injury litigation and the clinical oncologist.Clinical Oncology 10: 367–371, 1998

4. Hewitt HB, Wilson CW: The effect of tissue oxygen tensionon the radio-sensitivity of leukaemia cells irradiatedin situ inthe livers of leukaemic mice. Br J Cancer 13: 675–684, 1959

Timing of treatment 209

5. Fowler JF, Morgan RL, Wood CAP: Pre-therapeutic experi-ments with the fast neutron beam from the Medical ResearchCouncil cyclotron. 1. The biological and physical advantagesand problems of neutron therapy. Br J Radiol 36: 77–80, 1963

6. Sambrook, DK: Clinical trial of a modified (‘split-course’)technique of X-ray therapy in malignant tumours. Clin Radiol13: 1–18, 1962

7. Withers HR, Taylor JMG, Maciejewski B: The hazard of ac-celerated tumour clonogen repopulation during radiotherapy.Act Oncol 27: 131–146, 1988

8. Fowler JF, Lindstrom MJ: Loss of local control with prolong-ation in radiotherapy. Int J Radiation Oncology Biol Phys 23:457–467, 1992

9. Saunders MI, Dische S: Continuous, hyperfractionated, acel-erated radiotherapy (CHART). Sem Radiation Oncol 2: 41–44,1992

10. Johnson AE, Bennett MH, Cheung CWD, Cox SJ, Sales JELS:The management of individual breast cancers. The Breast 4:100–111, 1995

11. Cheung CWD, Johnson AE: Carcinoma of the breast: meas-urement and the management of treatment. II. The regressionof tumours. Br J Radiol 64: 121–132, 1991

12. Johnson AE, Shekhdar J: Interval cancers in the NHS breastscreening programme. Br J Radiol 68: 862–869, 1995

13. Orton, CG and Ellis F: A simplification in the use of the NSDconcept in practical radiotherapy. Br J Radiol 46: 529–537,1973

14. Fowler JF: How worthwhile are short schedules in radiother-apy? A series of exploratory calculations. Radiother Oncol 18:165–181, 1990

15. Fowler JF: The phantom of tumour treatment – continuallyrapid proliferation unmasked. Radiother Oncol 22: 156–158,1991

16. Fowler JF: Modelling altered fractionation schedules. Br JRadiol Suppl 24: 187–192, 1992

17. Cheung CWD, Johnson AE: Carcinoma of the breast: meas-urement and the management of treatment. I. The value of thedata. Br J Radiol 64: 29–36, 1991

18. Ashton-Key M, Campbell ID, Rew DA, Taylor I, StradlingR, Coddington R, Wilson GD: The histochemical evaluationof proliferation in breast carcinomas labelledin vivo withbromodeoxyuridine. The Breast 2: 42–47, 1993

19. Steel GG: Growth kinetics of tumours. Clarendon Press,Oxford 1977, pp 40–52

20. Johnson AE: Problems associated with randomized controlledclinical trials in breast cancer. J Eval Clin Pract 4: 119–126,1998

21. Johnson AE: Riposte to guest commentaries on ‘Problems as-sociated with randomized controlled clinical trials in breastcancer’. J Eval Clin Pract 4: 231–236, 1998

22. Bennett MH:In vivo labelling of human tumours with bro-modeoxyuridine: a histopathologist’s view. Br J Radiol Suppl24: 168–173, 1992

23. Wilson GD, Dische S, Saunders MI: Studies with bromod-eoxyuridine in head and neck cancer and accelerated radio-therapy. Radiother Oncol 36: 189–197, 1995

Address for offprints and correspondence:Ann Johnson, BreastStudy Centre, Mount Vernon and Watford Hospitals NHS Trust,Northwood, Middlesex, HA6 2RN, UK;Tel.: 01923 844502;Fax:01923 844258;E-mail: annjohnson@doctors.org.uk