Radiotherapy and Oncology 84 (2007) 5–10www.thegreenjournal.com

Waiting time

Tumor progression in waiting time for radiotherapyin head and neck cancer

Anni Ravnsbæk Jensena,b,*, Hanne Marie Nellemannc, Jens Overgaarda

aDepartment of Experimental Clinical Oncology, bDepartment of Oncology, and cDepartment of Radiology,Aarhus University Hospital, Denmark

Abstract

Introduction: Waiting-time prior to radiotherapy is a well-known problem. This study aims to determine the impact oftime on tumor growth in a patient population with squamous-cell carcinoma of the head and neck (SCCHN).Material and methods: In a consecutive cohort, all patients with both a diagnostic scan and a treatment-planning scan

were identified. In total 648 patients were seen, and 414 treated with primary radiotherapy. Ninety-five had two scansand 61 sets were eligible for comparison. Endpoints were change in tumor volume, tumor volume doubling time (TVD)and disease progression measured by TNM-classification and RECIST criteria.Results: Median interval between eligible scans was 28 (5–95) days. Thirty-eight (62%) had measurable increase in

tumor volume, median 46% (6–495%). For all patients TVD was median 99 days, but for the half of patients with fastestgrowing tumors TVD was 30 days (15–41). Tumor volume increase was significantly correlated to time and histologicaldifferentiation. Twelve (20%) developed new lymph-node metastasis and 10 (16%) progressed in TNM-classification.Evaluated by RECIST criteria 18 (30%) patients had progressive disease.Interpretation: This study shows a negative impact of waiting time in patients with SCCHN. Within an average time of 4

weeks the majority of the patients developed significant signs of tumor progression. It was not possible to define athreshold for acceptable time intervals in order to avoid volume changes, or to define a subgroup that has no negativeimpact of delay.

�c 2007 Elsevier Ireland Ltd. All rights reserved. Radiotherapy and Oncology 84 (2007) 5–10.

Keywords: Head and neck cancer; Delay; Waiting time; Radiotherapy; Tumor volume doubling time

Waiting time prior to radiotherapy is a major problem inmany radiotherapy centers with reported delays of 70 daysor even more [1,2]. Though frequently reported, it is stillnot an acceptable situation having the biology of cancer dis-eases in mind [3].

Studies analyzing the impact of waiting time on local con-trol probability and survival for patients with squamous cellcarcinoma of the head and neck (SCCHN) have shown variousresults [4–8]. This might partly be explained by a potentialrisk of selection bias such as giving priority to patients withadvanced tumor burden. To reduce bias, we analyzed theimpact of time on objective tumor parameters which areknown to affect prognosis. Thus a clear correlation betweentumor size and treatment response is known for patientswith head and neck cancers [9–11]. It is also well docu-mented that the local control probability is related to nodalstatus and to the volume of metastatic nodes [12–14].

This study aims to determine the impact of time betweena diagnostic and a treatment planning scan on tumor growthand on the development of new lymph node metastasis in apopulation based cohort of patients with SCCHN.

0167-8140/$ - see front matter �c 2007 Elsevier Ireland Ltd. All rights re

Materials and methodsAll patients with squamous cell carcinoma in pharynx,

larynx or oral cavity seen at the Department of Oncology,Aarhus University Hospital from January 2000 up to May2005 were identified. Altogether 648 patients were seenand 414 received primary radiotherapy with curative intent(Fig. 1). Ninety-five patients were identified, with both adiagnostic scan (MR or CT) and a treatment planning CT scan(Fig. 1). The final study group consisted of 61 patients withtwo sets of comparable scans, on which it was also possibleto make valid tumor measurements. For the remaining 34patients, tumors were either not measurable or patientswere not lying in comparable positions on both scans. Foreach case the diagnostic and the planning scan were evalu-ated successively in order to make measurements at thesame level and discriminate developing lymph nodes. Tumorsize and number and size of metastatic lymph nodes weremeasured by the same radiologist. Lymph nodes were con-sidered positive only if the largest diameter was more than1 cm. Endpoints were changes in tumor volume, tumor vol-ume doubling time, changes in TNM stage, and progressive

served. doi:10.1016/j.radonc.2007.04.001

648 patients initially seen at the department of radiotherapy

234 patients referred to Surgery/ palliation

414 patients had primary radiotherapy

124 patients without treatment planning scans

290 patients given primary radiotherapy with planning scans

195 patients without diagnostic scans

95 patients with a diagnostic and a treatment planning scan

61 patients eligible for the study

34 patients with scans not comparable or tumors not assessable

9 Larynx 17 Oral cavity 35 Pharynx

Fig. 1. Patient selection scheme.

6 Tumor progression in waiting time

disease measured by RECIST criteria [15]. TNM staging wasdone according to the UICC classification [16].

SPSS 13.0 statistical software for PC was used. Chi-square was used for testing differences in distribution offactors in n-by-n tables. The relationship between tumorvolume and waiting time was analyzed using Spearman‘sq. The impact of tumor differentiation, site and size wereanalyzed using partial correlation. All tests were two-tailed and a p-value of 0.05 or less was considered to bestatistically significant.

Tumor volume was estimated by calculating the volumeof an ellipse; whenever the third diameter was not avail-able, the mean of the two others was used. The third diam-eter was only used when measurable on both scans. Volumeincreases below 5% were regarded as within measurementerrors. Tumor volume doubling time (TVD) was calculatedby the formula: TVD = (ln2/ln (V2/V0)).

ResultsNinety-five patients had two scans performed and com-

pared to all the patients treated with curatively intendedradiotherapy these patients had more advanced disease,primarily because the group contained relatively lesspatients with small larynx tumors. Of the 95 sets of scans,it was possible to find comparable scans for 61 patients,and there were no differences between all the 95 patients

and the 61 patients with comparable scans. Eligible forthe study were 47 men and 14 women with a median timebetween the scans of 28 days (5–95). The composition ofthis group in relation to the whole patient cohort is shownin Table 1. For 45 patients the measured volume was pri-mary tumor and lymph node metastases and for 16 patientswhere tumor bed was not measurable, the measured vol-umes were from lymph node metastasis only. There wasno difference in growth rate between the two groups.

Overall, 38 patients (62%) showed a measurable increasein total tumor volume (primary tumor and regional lymphnodes) with a median volume increase of 46% (6–495),Fig. 2. The growth could be rather dramatic (Fig. 3), butthe outcome was heterogeneous and ranged from very fastgrowing tumors to tumors with no measurable growth. Thevolume increase was dependent on the scanning interval(Table 2, Fig. 2). Thus, among patients with an interval of2 weeks or less 4 of 12 patients showed significant progres-sion (33%), for the interval of 2–4 weeks 13 of 19 patientsshowed measurable progression (68%) and a similar fre-quency of 70% (21/30) was found for patients with morethan 4 weeks’ interval between scans.

In order to compensate for the difference in scanningintervals the tumor volume doubling time was estimatedusing data for the total volume of malignant lesions. Forall 61 tumors a median tumor volume doubling time of 99days (15 to >234) was observed. The TVD was independentof the initial tumor size (p = 0.81) and of the site of the pri-mary tumor (p = 0.58), but was found to be most pro-nounced in tumors with moderate to poorly differentiatedsquamous cell carcinomas (p = 0.003). The wide range indi-cated that about 38% had no measurable TVD whereas onthe other hand the quartile of patients with the highestgrowth rates had a median TVD of 27 days (15–35), andfor the half of patients with fastest growing tumors themedian TVD was 30 days (15–99).

In general, our patients had quite advanced tumors with23% being classified as T4, 58% were N2 or N3, and 62% instage IV at the time of the first scan. For patients withtumors less than T4 2/47 (5%) progressed to a larger T-clas-sification and 8/55 (15%) patients with less than N3 diseaseadditionally progressed to a larger N-classification resultingin 4 of 23 (17%) patients with less than stage IV tumors show-ing progression to a higher stage. Since this due to the initialadvanced stage may underestimate the true progression,this was also measured by use of the RESIST criteria whichare independent of actual stage. Using this parameter 18patients (30%) showed progressive disease (>20% increasein largest tumor diameter and/or appearance of new le-sions). Thus overall did 21 (34%) patients show increase inT or N-classification and/or RECIST progression (Table 2).

Both the progression in T, N, stage and the positive RE-SIST criteria were correlated with increasing time intervalbetween the scans.

DiscussionSeveral studies have analyzed the impact of patient and/

or professional delay on local control probability and overall

Table 1Patient characteristics

414 patients givenradiotherapy withcurative intent

290 patientswith treatmentplanning scan

61 patients withcomparable scans

Age (median) 60 (32–92) 58 (32–90) p < 0.01a 56 (32–81) p < 0.01b

Gender p = 0.54 p = 0.58Male 324 (79%) 230 (79%) 47 (77%)Female 89 (21%) 60 (21%) 14 (23%)

T-stage (%) p < 0.01 p < 0.01T1 115 (28%) 65 (22%) 10 (16%)T2 143 (35%) 109 (38%) 19 (31%)T3 107 (26%) 77 (27%) 15 (25%)T4 45 (11%) 36 (12%) 14 (23%)Tx 4 (1%) 4 (2%) 3a (5%)

N-stage p < 0.01 p < 0.01N0 230 (56%) 135 (46%) 13 (21%)N1 57 (13%) 45 (15%) 13 (21%)N2 104 (25%) 93 (32%) 29 (48%)N3 21 (5%) 19 (7%) 6 (10%)Nx 2 (1%) 1 (0%) 0

Stage p < 0.01 p < 0.01I 88 (21%) 41 (14%) 1 (1%)II 83 (20%) 58 (20%) 5 (9%)III 110 (27%) 77 (27%) 17 (28%)IV 131 (32%) 114 (39%) 38 (62%)

Site p < 0.01 p < 0.01Oral cavity 108 (26%) 92 (32%) 17 (28%)Larynx 198 (48%) 98 (34%) 9 (15%)Pharynx 108 (26%) 100 (34%) 35 (57%)

Type of diagnostic scanMR 34 (56%)CT 27 (44%)Days between scans (median) 28 (5–95)

The patient cohort compared with all having a treatment planning scan and patients eligible for the final study group.a All 414 patients vs. 290 with treatment planning scans.b All patients vs. 61 patients with two scans.

A. Ravnsbæk Jensen et al. / Radiotherapy and Oncology 84 (2007) 5–10 7

survival with conflicting results. These studies are all retro-spective and the interpretation is difficult due to randomvariations and confounding factors. A Danish study found asignificant longer waiting time for patients with smalltumors than for patients with large tumors [17], and it mightbe that patients with a history of fast growth or advancedstate might bypass the waiting list and receive more prompttreatment in centers struggling with capacity problems. Ithas been suggested that the aggressiveness of the disease it-self influences delay, a phenomenon called ‘‘the waitingtime paradox’’ [18].

In the current study we analyzed the impact of time onwell-established prognostic factors in a complete single-center patient cohort in order to avoid selection bias.Compared with the entire patient cohort, the eligible pa-tients included a higher proportion with locally advanceddisease, mainly because none of the patients with smallglottic cancers had two scans. It might introduce a bias,but we did not find any correlation between tumor sizeand growth and other studies suggest the same impact ofdelay in glottic cancer as in other sites in head and neck[7,8].

Sixteen percent progressed in T or N classification eventhough our patients had rather advanced stage at presen-

tation. For patients with locally advanced disease, theTNM classification is not sufficient to evaluate tumor pro-gression, as patients in stage IV cannot progress to a higherstage. Analyzing tumor growth according to the TNM clas-sification may thereby tend to underestimate diseaseprogression.

The mean increase in volume of the primary target lesion(tumor bed or the lymph node with the largest diameter atdiagnosis) was 38% in 28 days and is in accordance with theresults found by Waaijer et al., who found a mean increaseon 70% in 56 days [19]. Corresponding to these findings wealso found TVD within the same range being mean 87 (61–112) days in our study and mean 96 (26–256) days in thestudy by Waaijer (Fig. 4).

The consequences of delay in start of radiotherapy arerelated to TVD and for a patient with a fast-growing tumoran interval of less than one month can have a significantadverse effect on the outcome [20]. Mackillop et al. haveestimated an 8% loss in local tumor control probability by a4 weeks delay for tumors with a TVD on 90 days [21]. Ourstudy confirms that it seems to be a valid estimate of TVD.However the growth rates do not follow a Gaussian distri-bution and we found a large heterogeneity in growth rateswith TVD ranging from 15 days to tumors with no measur-

Pro

gres

sion

acc

ordi

ng to

RE

CIS

T

0

20

40

60

80

Time (days)

0

Incr

ease

in to

tal v

olum

e %

0

50

100

150

200

250

450

500

Total volume at diagnosis (cm3)

Tum

our

volu

me

doub

ling

time

0

50

100

150

200

250

20 40 60 80 100

Time (days)0 20 40 60 80 100

0 20 40 60 80 100 120

a

b

c

Fig. 2. Volume increase while waiting. (a) Increase in malignanttumor volume. 62% had measurable increase in volume with medianof 46%. (s) Highly differentiated tumors, (d) Poorly/moderatelydifferentiated tumors p < 0.01 for time vs. growth, p < 0.01 forcorrelation to histological differentiation. (b) Tumor volume dou-bling time as a function of malignant tumor volume at diagnosis,p > 0.05. (c) Progression according to RECIST. (d) Change in largesttumor diameter%, (h) Appearance of new lesions. Patients abovethe dotted line have progression according to RECIST criteria.

8 Tumor progression in waiting time

able growth. Assuming an initial tumor control rate of 50%and using the data from Mackillop et al. [21] we did esti-

mate quite marked reductions in tumor control probability.For the quartile of patients with the highest growth rates(TVD < 35 days), a more than 20% loss in local tumor con-trol probability could be expected within 4 weeks time(from 50% to approx. 30%), and for the half of patientswith the fastest growing tumors (TVD < 99 days) the esti-mated loss in local control probability would be more than8% within 4 weeks (from 50% to 42%) [21]. The impact onsurvival of such decrease in local control probability maybe substantial. Thus previous overviews have indicatedthat about 7 out of 10 loco-regional failures after radio-therapy for SCCHN may eventually die due to lack of tumorcontrol [22,23]. It should not be forgotten that the presentnumbers are model calculations and estimates. However,they are founded in well-described clinical studies andthere are no obvious arguments against their clinicalvalidity.

As described did we observe a large range between fast-and slow-growing tumors, also among tumors from the samesite and of the same size. It was not possible to distinguishthe very fast-growing tumors from the remaining except byhistological differentiation and it is therefore even lessacceptable with long time intervals for patients with poorlyor moderately differentiated tumors. However, even amongthe well-differentiated tumors did 4 of 13 (31%) show ameasurable increase in tumor volume.

In this study we are only dealing with a smaller part ofthe total time between diagnosis and start of treatment.In 2002 the median time interval for patients with cancerin head and neck in Denmark was 70 days for this patientgroup, a 3 weeks increase from the time observed in 1992[1]. According to the model by Mackillop [21] and demon-strated above, it corresponds to a substantial loss in localcontrol probability even assuming the most conservativeestimates of TVD.

The treatment outcome after radiotherapy for SCCHN isconstantly improving. In Denmark we have developed ourtreatment strategy based on a series of large national ran-domized clinical trials [24,25]. As an example we haveachieved an approximately 10% gain in tumor control byusing moderate accelerating fractionation schemes withsix fractions peer week [25,26]. The large trial by which thistreatment principle was derived [24] to place in the 1990s,and it is a bitter irony that while we were developing therationale for a biological based treatment improvement,this gain or even more may have been lost in the same timeperiod due to the above-described increase in time betweendiagnosis and start of therapy [1].

Other factors may additional aggravate the situation. Wecompared initial MR scans to later CT scans and it might beone of the reasons that so many scans were not comparable.However, when comparing the two modalities in this order,we might have underestimated tumor growth, as there is atendency to overestimate tumor size on MR compared toCT [27].

Also the observed increase in tumor volume, and theappearance of new lymph node metastasis within the sameT and N stage do not only affect prognosis but also increaseradiotherapy field sizes and thereby the frequency andseverity of early and late morbidity caused by the volumeeffect [28,29].

Fig. 3. Examples of tumor progression. (a) N-stage. A patient with 47 days in between scans. Note growth of the lymph node metastasis andappearance of a contra-lateral metastasis. Scans are within the same level but with differences in positioning of the chin. (b) T-stage. Apatient with 19 days between scans. Note the progressive bone destruction.

Table 2Changes in volume and RECIST criteria according to time interval between scans

Time interval (days) Number ofpatients

Increase involume

Change in T orN classificationa

Progressionaccording to RECIST

Change in RECISTand/or T or N classification

Any sign ofprogression

0–14 12 4 (33%) 0 0 0 4 (33%)15–28 19 13 (68%) 4 (21%) 6 (32%) 7 (37%) 13 (68%)>28 30 21 (70%) 6 (20%) 12 (40%) 14 (47%) 21 (70%)All 61 38 (62%) 10 (16%) 18 (20%) 21 (34%) 38 (62%)

a Note that several patients had initially tumors classified as T4 and/or N3.

A. Ravnsbæk Jensen et al. / Radiotherapy and Oncology 84 (2007) 5–10 9

ConclusionThis study shows a negative impact of waiting time in a sub-

stantial part of patients with SCCHN. With a median of 4 weeksintervalbetween twoscansdid themajorityofpatientsdevelopsignificant increase in measurable parameters for tumor vol-ume or progression. It was not possible to define a threshold

for acceptable time intervals in order to avoid volume changes,or to define a subgroup that has no negative impact of delay.

* Corresponding author. Anni Ravnsbæk Jensen, Department ofExperimental Clinical Oncology, Aarhus University Hospital, Nørre-brogade 44, Building 5, DK-8000 Aarhus C, Denmark. E-mailaddress: anni@oncology.dk

Vol

ume

incr

ease

%

0

50

100

150

200

250450

500

Time (days)0 20 40 60 80 100

Fig. 4. Volume increase while waiting including other known data ontumor growth. (h) Data from Waaijer et al. [19] with a meanincrease in tumor volume of 70% in 56 days. (d) Data from currentstudy with mean increase in tumor volume of 38% in 28 days.

10 Tumor progression in waiting time

Received 4 January 2007; received in revised form 1 April 2007;accepted 4 April 2007; Available online 9 May 2007

References[1] Primdahl H, Nielsen AL, Larsen S, Andersen E, Ipsen M, Lajer C,

et al. Changes from 1992 to 2002 in the pretreatment delayfor patients with squamous cell carcinoma of larynx orpharynx: a Danish nationwide survey from DAHANCA. ActaOncol 2006;45:156–61.

[2] Robinson D, Massey T, Davies E, Jack RH, Sehgal A, Moller H.Waiting times for radiotherapy: variation over time andbetween cancer networks in southeast England. Br J Cancer2005;92:1201–8.

[3] Rutqvist LE. Waiting times for cancer patients – a ‘‘slipperyslope’’ in oncology. Acta Oncol 2006;45:121–3.

[4] Barton MB, Morgan G, Smee R, Tiver KW, Hamilton C, Gebski V.Does waiting time affect the outcome of larynx cancer treatedby radiotherapy? Radiother Oncol 1997;44:137–1341.

[5] Fortin A, Bairati I, Albert M, Moore L, Allard J, Couture C.Effect of treatment delay on outcome of patients with early-stage head-and-neck carcinoma receiving radical radiother-apy. Int J Radiat Oncol Biol Phys 2002;52:929–36.

[6] Huang J, BarberaL, BrouwersM,BrowmanG,MackillopWJ.Doesdelay in starting treatment affect the outcomes of radiother-apy? A systematic review. J Clin Oncol 2003;21:555–63.

[7] Hansen O, Larsen S, Bastholt L, Godballe C, Jorgensen KE.Duration of symptoms: impact on outcome of radiotherapy inglottic cancer patients. Int J Radiat Oncol Biol Phys2005;61:789–94.

[8] Brouha XD, Op De CB, Terhaard CH, Hordijk GJ. Does waitingtime for radiotherapy affect local control of T1N0M0 glotticlaryngeal carcinoma? Clin Otolaryngol Allied Sci 2000;25:215–8.

[9] Overgaard J, Hansen HS, Jorgensen K, Hjelm HM. Primaryradiotherapy of larynx and pharynx carcinoma–an analysis ofsome factors influencing local control and survival. Int J RadiatOncol Biol Phys 1986;12:515–21.

[10] Bentzen SM, Johansen LV, Overgaard J, Thames HD. Clinicalradiobiology of squamous cell carcinoma of the oropharynx.Int J Radiat Oncol Biol Phys 1991;20:1197–206.

[11] Bentzen SM, Thames HD. Tumor volume and local controlprobability: clinical data and radiobiological interpretations.Int J Radiat Oncol Biol Phys 1996;36:247–51.

[12] Johansen LV, Grau C, Overgaard J. Glottic carcinoma–patternsof failure and salvage treatment after curative radiotherapy in861 consecutive patients. Radiother Oncol 2002;63:257–67.

[13] Lindeloev B, Hansen HS. The impact of lymph node metastaseson the results of treatment by primary radiotherapy andsecondary surgery in oropharyngeal cancer. Acta Oncol1995;34:965–8.

[14] Vergeer MR, Doornaert P, Leemans CR, Buter J, Slotman BJ,Langendijk JA. Control of nodal metastases in squamous cellhead and neck cancer treated by radiation therapy or chemo-radiation. Radiother Oncol 2006;79:39–44.

[15] Therasse P, Arbuck SG, Eisenhauer EA, Wanders J, Kaplan RS,Rubinstein L, et al. New guidelines to evaluate the response totreatment in solid tumors. European Organization for Researchand Treatment of Cancer, National Cancer Institute of theUnited States, National Cancer Institute of Canada. J NatCancer Inst 2000;92:205–16.

[16] UICC, TNM classification of Malignant Tumours, 5th edition.1997.

[17] Wildt J, Bundgaard T, Bentzen SM. Delay in the diagnosis oforal squamous cell carcinoma. Clin Otolaryngol 1995;20:21–5.

[18] Crawford SC, Davis JA, Siddiqui NA, de CL, Gillis CR, Hole D,et al. The waiting time paradox: population based retrospec-tive study of treatment delay and survival of women withendometrial cancer in Scotland. BMJ 2002;325:196.

[19] Waaijer A, Terhaard CH, Dehnad H, Hordijk GJ, van LeeuwenMS, Raaymakers CP, et al. Waiting times for radiotherapy:consequences of volume increase for the TCP in oropharyngealcarcinoma. Radiother Oncol 2003;66:271–6.

[20] Wyatt RM, Beddoe AH, Dale RG. The effects of delays inradiotherapy treatment on tumour control. Phys Med Biol2003;48:139–55.

[21] Mackillop WJ, Bates JH, O‘Sullivan B, Withers HR. The effectof delay in treatment on local control by radiotherapy. Int JRadiat Oncol Biol Phys 1996;34:243–50.

[22] Overgaard J, Horsman MR. Modification of hypoxia-inducedradioresistance in tumors by the use of oxygen and sensitizers.Sem Rad Oncol 1996;6:10–21.

[23] Wadsley JC, Bentzen SM. Investigation of relationship betweenchange in locoregional control and change in overall survival inrandomized controlled trials of modified radiotherapy in head-and-neck cancer. Int J Radiat Oncol Biol Phys 2004;60:1405–9.

[24] Overgaard J, Hansen HS, Overgaard M, Bastholt L, BerthelsenA, Specht L, et al. A randomized double-blind phase III studyof nimorazole as a hypoxic radiosensitizer of primary radio-therapy in supraglottic larynx and pharynx carcinoma. Resultsof the Danish Head and Neck Cancer Study (DAHANCA) Protocol5–85. Radiother Oncol 1998;46:135–46.

[25] Overgaard J, Hansen HS, Specht L, Overgaard M, Grau C,Andersen E, et al. Five compared with six fractions per weekof conventional radiotherapy of squamous-cell carcinoma ofhead and neck: DAHANCA 6 and 7 randomised controlled trial.Lancet 2003;362:933–40.

[26] Bourhis J, Overgaard J, Audry H, Ang KK, Saunders M, BernierJ, et al. Hyperfractionated or accelerated radiotherapy inhead and neck cancer: a meta-analysis. Lancet2006;368:843–54.

[27] Dammann F, Horger M, Mueller-Berg M, Schlemmer H, ClaussenCD, Hoffman J, et al. Rational diagnosis of squamous cellcarcinoma of the head and neck region: comparative evaluationofCT,MRI, and18FDGPET.AmJRoentgenol 2005;184:1326–31.

[28] Withers HR, Taylor JM, Maciejewski B. Treatment volume andtissue tolerance. Int J Radiat Oncol Biol Phys 1988;14:751–9.

[29] Bentzen SM, Saunders MI, Dische S, Bond SJ. Radiotherapy-related early morbidity in head and neck cancer: quantitativeclinical radiobiology as deduced from the CHART trial. Radio-ther Oncol 2001;60:123–35.