Radiotherapy and Oncology, 1 (1983) 53-63 Elsevier

RTO 00009

53

Causes of failure of radiotherapy in head and neck cancer

Lester J. Peters and Gi lbe r t H. F le t che r

Division of Radiotherapy, The University of Texas M. D. Anderson Hospital and Tumor Institute at Houston, 6723 Bermer Avenue, Houston, TX 77030, U.S.A.

(Received 11 May 1983, accepted 24 May 1983)

Keywords: Head and neck cancer; Failure; Radiotherapy

Summary

Causes of failure of radiation therapy are reviewed and illustrated with clinical examples from cancers of the head and neck region. Radiobiological factors relating to volume of cancer, hypoxia, tumor cell kinetics, intrinsic cellular radiosensitivity and repair capability are considered, along with physical factors relating to fraction size and inadvertent underdosage. In addition, reference is made to failures attributable to a sigmoid dose response curve for tumor control and to the development of second primary cancers. The distinction is made between those causes of failure that can be minimized by optimal application of concepts and techniques readily available to all radiotherapists, those that are not amenable to any modification of ra- diotherapeutic technique, and those that are potentially remediable by new treatment strategies based on the radiobiological attributes of individual tumors.

Introduction

Analysis of causes of failure of radiotherapy serves two important functions. It suggests ways in which failures may be reduced by appropriate modifica- tion of radiotherapy technique, or when this is not possible, it delineates the circumstances under which radiotherapy should not be used for primary treatment. Cancers of the head and neck region lend themselves to accurate evaluation of the causes of failure because they are generally easier to stage and to follow than tumors in other sites of the body. However, the conclusion of analyses reached on the basis of head and neck cancer can be extrapolated to other anatomical sites. We will review causes of

failure not only for definitive radiation therapy, but also for the widely used approach of combined treatment with surgery and post-operative radia- tion therapy.

Histology

For the first half of the century, radiosensitivity was considered to be a simple function of histology, Different histological tumor types were classified as radiosensitive, moderately radiosensitive, or radio- resistant [17], and for the radioresistant tumors there was no need to look for causes of failure other than their histology. Furthermore, within the mod-

0167-8140/83/$03.00 �9 1983 Elsevier Science Publishers B.V.

54

erately radiosensitive group, differences in radio- sensitivity were attributed to anatomical location. For instance, squamous carcinoma of the mobile tongue was considered to be more radiosensitive than squamous carcinoma in the base of the tongue, or squamous carcinoma metastatic to cervical lymph nodes. This conclusion can be traced to the fact that higher doses of radiation could be deliv- ered using interstitial techniques to the oral tongue compared with the limited doses that were feasible with external beam therapy using kilovoltage treat- ment machines.

Although it is undoubtedly true that tumors of certain histological types are more radiocurable than others, it is by no means established that this correlates with inherent differences in cellular ra- diosensitivity (see below), as opposed to epigenetic and environmental factors such as the proport ion of tumor stem cells, tumor cell kinetics, and oxygen status. In fact, the available evidence suggests that histology per se is not a major determinant of ra- diocurability. This conclusion can be reached from

TABLE I

Local-regional failures by histologic types after resection and postoperative irradiation.

Histology No. of Failures patients

Primary Neck Primary site a site

and neck

Malignant mixed 13 1 0 0 Adenocarcinoma 17 2 2 1 Mucoepidermoid 9 0 0 0

(low grade) Mucoepidermoid 15 1 2 0

(high grade) Acinic cell 5 0 0 0 Adenoid cystic 15 1 0 0 Unclassified 1 0 0 0 Other 2 0 0 0

the fact that the control of subclinical disease in various anatomical sites is essentially independent of histology [12]. An example from the head and neck region is provided in Table I which shows that for a variety of histological types of parotid cancer treated by surgery and postoperative irradiation, failures occurred with approximately equal fre- quency among the more malignant histotypes.

Volume o f cancer

The probability of control of any cancer at a given dose level is a function of the number of clonogenic cells that need to be eliminated. It is not possible to measure directly the number of clonogenic cells in human tumors, but for different classes of tumors, it is reasonable to suppose that the clonogenic cell number is closely correlated with tumor volume. The influence of volume of cancer on the probabil- ity of control of cervical lymph node metastases from squamous carcinomas of the laryng0pharynx is illustrated in Table II. These data show not only a decreasing control probability with increasing size of lymph nodes within each dose range, but also a dose dependence for control within each size range.

The principle of volume of cancer also has important application in the treatment of subclini- cal disease. The concept of subclinical disease as advanced by Fletcher [I 1] is precise and was applied to specific anatomical areas amenable to clinical examination. The two main applications were as follows: (1) Clinically negative lymphatic areas in

TABLE II

Control rates a as a function of the size of the node(s) and radia- tion dose in the squamous cell carcinomas of the laryngo- pharynx.

Dose Size Dose Size

< 3 c m 3 5cm > 5 c m

_<65Gy 15/26 (58%) _<70Gy 3/9 (33%) 1/9 (11%) > 6 5 G y 86/95 (91%) > 7 0 G y 11/13 (85%) 11/15 (73%)

Total 77 5 4 1

" Including the facial nerve into the base of the skull. From McNaney, D. et al.: Int. J. Radiat. Oncol. Biol. Phys., in press, 1983.

" Determinate group. From Bataini et al [4].

55

which a high probability of occult disease was known to exist: (2) areas rendered free of clinically detectable disease by limited surgical excision of a circumscribed tumor mass. In these situations it has been repeatedly substantiated that 50 Gy in 25 frac- tions over 5 weeks eradicates in excess of 95% of occult disease.

However, for conventional postoperative therapy the concept of subclinical disease is not directly ap- plicable, and the dose of radiation required must be modified according to the density of infestation of the operative bed by tumor cells as deduced from the surgical pathology. Another consideration is that after extensive surgical dissections, residual tu- mor cells are more likely to be in a poorly vascu- larized environment. This can frequently be ap- preciated clinically by a lesser skin reaction and/or failure of epilation in a surgical flap when compared with adjacent undisturbed skin in the radiation field. For these reasons there is no single entity of subclinical disease in the postoperative setting, and the dose required must be based on the features of the individual case.

Tumor cell hypoxia

Over the past three decades, hypoxic cells have been considered the single most significant cause of treat- ment failure, and most new radiotherapy regimens are still predicated on this belief. Although the ex- istence of hypoxic tumor cells is indisputable, the question of how frequently they determine treat- ment failure with optimum fractionated techniques is still the subject of controversy. The most direct evidence would be expected from trials of radio- therapy in hyperbaric oxygen versus ambient con- ditions, or from trials of selective hypoxic cell ra- diosensitizers. However, with very few exceptions, these trials have shown benefit only when a few high dose fraction treatment regimens have been em- ployed [8]. This may be related to the greater influ- ence of a biphasic survival curve when large dose fractions are used, and also to the probability that the oxygen enhancement ratio is lower with small doses per fraction where cell killing is mainly due

to the single-hit or alpha-component [1]. Further support for this interpretation is obtained from the success of low dose rate interstitial techniques in the head and neck region.

Deficient reoxygenation is a potential cause of failure of fractionated treatment. However, it should be remembered that reoxygenation need not be complete when using fractionated treatment, and, in fact, with incremental doses of 200 cGy or less the influence of hypoxic cells would not be sig- nificant unless they constituted close to 50% of the viable tumor cell population at the time of each dose [26].

Tumor regeneration

Because of the relatively slow growth rate of most human tumors (median volume doubling time about 2 months) [6], tumor cell regeneration during a conventional course of fractionated treatment lasting 6-7 weeks has generally been discounted as a cause of failure. However, the critical parameter

Pyriform Sinus Cumulative Rate of Failures Above the Clavicles

1.00

.80

.60

AO

.20 f i

Total Failures

14 o P o s t - O p X R T P t $ 8 0 ~ S u r g e r y O n l y P t e

i i i i i i i

0 12 24 36 48 Months

Fig. l. Cumulative appearance rates of failures above the cla- vicles. Almost half of the failures in the surgery only group have appeared by 6 months. From Fletcher [14].

56

TABLE III

Recurrences in the head and neck of patients electively irradiated postoperatively for Stages Ill and 1V epidermoid carcinoma of the head and neck.

Delay up to 6 wks Delay over 6 wks"

No. (%) No. (%)

Nodes negative 0/5 0 0/3 0

Nodes positive at one level 0/18 0 3/11 27

Nodes positive at multiple levels 2/25 8 12/32 37

Total 2/48 4 l 5/46 32.5

a In the patients with a delay over 6 weeks there was no palpable disease at the time of the initiation of irradiation.

From Vikram et al. [23].

is not the volume doubling time, but the potential doubling time of the clonogenic tumor cell popu- lation. The most direct evidence to implicate tumor cell regeneration as a cause of failure in head and neck cancer comes from the data of Parsons et al. [16]. These authors showed that when the same to- tal dose in the same number of fractions was given as either a continuous or split course, local control was superior in all stages for the continuous treat- ment. This implies that regeneration occurring dur- ing the 2 weeks gap in the split course regimen

contributed to tumor failure. The rapid rate of proliferation of tumor cells sur-

viving unsuccessful radiotherapy is attested to by the fact that the great majority of failures in head and neck cancer are manifest within 2 years of treat- ment regardless of tumor growth rate prior to treat- ment (Fig. 1). Furthermore, delay between surgery and the initiation of postoperative radiotherapy has been shown to have a major influence on the success of the combined treatment with a delay of 6 weeks or more being associated with a much higher re- lapse rate than with shorter time intervals (Table

III). In certain circumstances, even the continuous ad-

ministration of lO Gy per week may not be suffi-

cient to counter tumor regeneration. For example, it is not uncommon for neck metastasis to grow during treatment and occasionally patients who had no palpable disease at the initiation of post- operative irradiation can develop overt disease dur- ing t reatment (Fig. 2). In these circumstances ac- celerated fractionation in the form of a concomitant boost is indicated [22].

Failures of redistribution

Failure of redistribution is another potential cause of failure of conventional fractionated treatment. It is well documented that significant variations in cellular radiosensitivity exist as a function of cell age in the division cycle. In fact, with small incre-

mental doses the magnitude of differences in radio- sensitivity attr ibutable to the cell age can exceed that imposed by tumor cell hypoxia [20]. It follows that tumors having a large non-growth fraction of clonogenic cells would therefore be relatively re- sistant to conventional fractionated t reatment be- cause of nonprogression of cells from more resist- ant phases of the division cycle.

A long tumor cell turnover time is also correlated with a slow volume regression after commencement of radiotherapy, and it is possible that failure of redistribution could account for the clinical obser- vation that tumors which do not "clear" completely during a course of t reatment are less likely to be cured [3]. However, other factors, such as failure of reoxygenation cannot be excluded. Since variat ion in radiosensitivity with cell age is much less with high LET radiations, it was predicted [25] that the RBE of tumors that redistributed poorly would be high, providing a "kinetic gain factor" for neutron therapy. Limited clinical confirmation of this pre- diction has been provided by studies on lung me- tastases with widely varying growth rates [5] and kinetic analysis of tumors to select tumors with a low turnover rate for trials of high LET irradiations would appear rational and justified.

57

Fig. 2. Patient was referred in June 1979 for postoperative irradiation after extensive surgery for a squamous cell carcinoma of the ear. At the start of irradiation there was no palpable disease in the entire area. Approximately one week after the start of irradiation at 10 Gy/week, a suspicious area was paplated in the supraclavicular area; by the second week it was definite. Through a small field within the large field 5 x 200 cGy was added in one week (A). The second treatment was given as a concomitant boost 3 h after the first one. At the completion of treatment there was moist desquamation of the skin in the field-within-the-field area (B). The total dose to the nodule was 66 Gy in 5 weeks. The tumor had clinically disappeared and the patient was NED in May 1982. From Fletcher [14].

Intrinsic cellular radiosensitivity and repair capa- bility

A l t h o u g h no di rec t evidence can be a d d u c e d to

show tha t the intr insic ce l lu lar r ad iosens i t iv i ty and /

or repa i r capab i l i t y o f t u m o r cells is a m a j o r cause

o f fai lure in the t r ea tmen t of h u m a n cancer , r ad io -

b io log ica l d a t a suggest t ha t this is l ikely to be the

case. Signif icant differences in the ini t ia l s lope and

shape o f the shou lde r o f the survival curve o f var-

ious h u m a n t u m o r cell lines have been d e m o n s t r a -

ted [9] and re la t ively small differences in cell sur-

vival af ter low doses wou ld have an e n o r m o u s effect

on the success o f a f r a c t i ona t ed course o f t r e a t m e n t

due to the ampl i f i ca t ion o f the f r ac t iona t ion p ro -

cess. F o r example , the difference be tween a surviv-

ing f rac t ion o f 60% or 50% af ter an inc remen ta l

dose o f 200 c G y w o u l d lead to a difference o f 2.8

logs o f cell k i l l ing in a f r ac t i ona t ed course o f 70 Gy ,

a s suming equal effect per f ract ion.

However , the t rue m a g n i t u d e o f differences in

survival f rac t ion af ter each dose m a y no t be repre-

sented in classical cell survival exper iments . Wei -

chse lbaum and Li t t le [24] have recent ly reviewed

the differences in capac i ty to repa i r po t en t i a l l y

le thal d a m a g e be tween va r ious cell lines der ived

f rom h u m a n tumors . These a u tho r s m a k e the case

tha t such differences m a y be more i m p o r t a n t t han

classical cell survival in de t e rmin ing the o u t c o m e o f

t r ea tmen t . This a rea is c lear ly one deserv ing o f re-

search emphas i s , especial ly wi th r ega rd to measu re -

ments on fresh h u m a n t u m o r explants .

58

Fraction size

Extensive clinical experience supports the observa- tion that use of large fractional doses changes the proportionality between acute and late normal tis- sue reactions with a relative increase in severity of late reactions [22]. Insofar as tumors can be regard- ed kinetically as more resembling acutely reacting normal tissues than late reacting ones, it can be ar- gued that use of small incremental dose fractions will increase the therapeutic ratio between tumor control and late normal tissue injury. This is the rationale for hyperfractionation. Using the same argument, one would predict that the use of larger dose fractions would be associated with a poorer therapeutic ratio. This prediction is supported by the analysis of results by Cox et al. [7] showing that

hypofractionated regimens with the same NSD were associated with poorer local disease control than conventional fractionation schemes. However, conflicting results have been reported by Henk [15] who observed no difference in disease control in a randomized trial using 30 or 10 fractions for de- finitive treatment of head and neck cancer.

Sigmoid dose-response curve

As a consequence of the randomness of cell killing by radiation, the probability of curing any tumor is determined by Poisson statistics and is a sigmoid function of dose. Up to a certain dose the tumor control probability is small, following which there is a steep increase with increasing dose until a pla-

2T3 o

~ W

Pharyngeal Wall

T1 + T 2

76% ~37/49)

T 3 ~, T 4

46% (45/98)

I, 25 ' ' '

5 6 7 8 Equivalent Dose

Retromolar Trigone/Anterior Faucial Pillar

25

T 1 + T 2

i i

s ; ; 8 T 3 + T 4

84% (38/45)

Tonsillar Fossa

T 1 + T 2

84% (32/38)

T 3 + T 4

63% (24138)

, I | I I

5 6 7 8 (Kilorad)

Supraglottic Larynx

T 2 + T 3

7

I I I I

5 6 7 8 Equivalent Dose (Kilorad)

Base of Tongue I

T 1 + T 2 [

25

75

M

25

T 3 + 3" 4

59% (41/69)

5 6 7 8 Equivalent Dose (Kilorad)

Fig. 3. Dose-response curves for two-stage groupings of squamous cell carcinomas arising in various head and neck sites. In each case, the data were analyzed as one group using logistic regression and Poisson models to fit tumor control probability as a function of tumor dose expressed as a 30 fraction, 6 weeks equivalent. Significantly positive dose-response relationships exist for the T1 + T2 retromolar trigone-anterior faucial pillar and T2 + T3 supraglottic larynx grouoings. From Thames et al. [21].

teau is reached in which a large additional dose is necessary to achieve a relatively small increase in tumor control probability. A great deal of spurious argument exists in the literature concerning the presence or absence of dose-response curves for the control of head and neck cancers. Unless one pos- tulates that there are subpopulat ions of tumor cells which are absolutely resistant to radiation (which has never been observed), it is impossible to escape the conclusion that tumor control probabil i ty is a Poisson function of dose. The problem of demon- strating dose-response relationships in clinical ma- terial is at tr ibutable to the following factors:

(1) H u m a n tumors, even within the same stage category, are heterogeneous with regard to all of the features determining radiocurability. Such het- erogeneity inevitably acts to flatten the dose-re- sponse relationship [19].

(2) Patients have been randomly assigned to re- ceive different dose levels in very few clinical stud- ies. Thus, the bias exists to give higher doses to patients with more advanced disease which again will tend to flatten the dose-response function.

(3) Rarely are doses calculated at the site of tu- mor failure and this can differ significantly from the

prescribed tumor dose. (4) With few exceptions, no quality assurance

can be obtained that technical errors did not con- tribute randomly to causes of failure.

In spite of these limitations, at least some clinical dose-reponse analyses show significant positive slopes (Fig. 3). Even with steep curves, however, there is a point beyond which the likelihood of im- proving the therapeutic ratio is small. The more heterogeneous the populat ion the lower the tumor control probabil i ty at which this plateau is reached.

59

flagrant and obvious geographic misses still unfor- tunately occur, we will focus here on subtle and sometimes unappreciated examples of underdos- age. Examples from the head and neck region in- clude the following.

Off-axis dosimetry. Inadvertent underdosage may account for the cause of failure of at least 10% of cancers of the nasopharynx. This is because the pri- mary tumor is located in the corner of the radiation field and the actual dose received is significantly less than the calculated central axis dose because the isodose curves are not flat to the geometric edges of the beam. Although less significant than with kilovoltage irradiation, differential bone absorpt ion

17 MeV 7.3 cm x 6 . 8 cm 100 SSD

Inadvertent or underdosage

The surest cause of radiotherapy failure is the geo- graphic miss or inadvertent underdosage. This fac- tor must always be excluded in any analysis of dose- reponse relationships for radiotherapy. Although

Fig. 4. Isodose distribution, with two dimensional corrections for tissue inhomogeneities for a direct anterior 17 MeV electron beam field to the nose. Doses along the beam axis are signifi- cantly lower than in a homogeneous phantom (e.g. at the ex- pected 90% depth of 5.2 cm, the dose is in fact 64%). On the other hand, serious hot spots exist off the beam axis. From Fields and Hogstrom [10].

6O

TABLE IV

Control of primary lesion correlated with T stage squamous cell carcinoma of retromolar trigone-anterior faucial pillar megavoltage treatment alone - March 1954 through August 1973 (analysis, November 1975).

Stage Total P r imary Definite ultimate Underdose or NED NED locally Analyzed failure failure after geographic 2 yrs expired 2 yrs case

surgical salvage miss other causes (DM, ID, UNK)

T1 26 15 % 4% 2 18 4 20 (4/26) (1/26)

T2 103 19% 7% 5 66 17 81 (20/103) (7/103)

T3 61 15% 3% 5 36 16 40 (9/61) (2/61)

T4 14 36% 21% 2 2 7 5 (5/14) (3/14)

Abbreviations: DM = distant metastases; ID = intercurrent disease; UNK = unknown cause; NED = no evidence of disease. From Barker and Fletcher [2].

also contributes a little to the underdosage of the primary tumor. An analysis of nasopharyngeal tu-

mors treated at U .T .M.D.A.H. from 1954 to 1971 with a nominal 60 Gy tumor dose for T2 squamous cell carcinomas showed a significant incidence of failure at the primary site (12/43). Subsequent to 1972, a small field centered over the angle of the nasopharynx was used to deliver an extra dose of 50(~750 cGy, and this reduced the incidence of fail-

ure of T2 squamous cell carcinoma to 1/14 [13]. This cause of failure by underdosage was not evi- dent with the lymphoepitheliomas since lower doses

are required for the same tumor control probability as with "pure" squamous cell carcinoma.

Skin sparing. The increasing use of high energy photons for head and neck cancer has led to the risk of inappropriate skin sparing, especially in the postoperative setting. In general, photon energies in excess of 4 MeV or 6~ are unwise when the

neck is being treated postoperatively. A related but less well known effect is the build-down pheno- menon. This refers to the reduction of dose at a

tissue-air interface as a photon beam exits the tissue due to lack of backscatter. Thus large cavities, for

example after a radial maxillectomy, should be

treated with the obturator in place or other suitable bolus.

Treatment with electrons or a mixture of electrons and high energy photons. Electron beams are sub- ject to a much greater risk of inadvertent under-

dosage than photons. This is because of the greater perturbation introduced by tissue inhomogeneities,

and the possibility of a geographic miss in depth. Figure 4 shows the isodose distribution obtained from a direct anterior electron beam field to the region of the nose with correction for tissue inhom- ogeneity. It is readily apparent that a major under- dosage can result if these corrections are not made. Figure 4 also shows that significant hot spots can coexist with tumor underdosage if simple uncom- pensated electron beam portals are used. Another dosimetric problem shared by electron and high en- ergy photon beams is forward peaking of the iso- dose curves to a much greater extent than that as- sociated with 4-6 MeV X-rays or well-trimmed 6~ beams. This is illustrated by an analysis of

failures in squamous cell carcinomas of the retrom- olar trigone-anterior faucial pillar treated at U.T. M.D.A.H. Table IV shows that the failure rate is about the same as one progresses from T1 to T3 on

linear dimensions reflecting the small change in vol- ume of cancer. However, in all stages most failures occurred anteriorly. Analysis showed that patients treated with parallel opposed 6~ fields with equal or 2:1 loading had fewer failures than those treated with ipsilateral high energy electron beams alone or in combinat ion with high energy photons. In retro- spect, this can be ascribed to the significant forward peaking of the isodose curves of these beams. Le- sions of the re t romolar trigone-anterior faucial pil- lar often do not have a clear delineation, being sur- rounded by areas of leukoplakia along the buccal mucosa and lower gum. Thus, microscopic disease anteriorly is easily underdosed. Subsequent to the analysis of Table IV, the use of a generous anterior margin has significantly diminished the incidence of failures in this location.

Second primary cancers

Apparent failure of radiotherapy can result f rom the development of a second pr imary cancer within or adjacent to the previously treated area. This is particularly relevant to the head and neck region,

61

since the habits contributing to the development of cancers are usually perpetrated by this patient population. In some circumstances, it is difficult to distinguish between a marginal recurrence due to inadequate coverage, as described above in relation to cancer of the re t romolar trigone-anterior faucial pillar, and genuine second pr imary cancers occur- ring within the irradiated volume. However, a long time to "recurrence" after apparently successful t reatment points toward the development of a new pr imary lesion. The data in Table V illustrate the point with regard to cancers of the larynx. Whereas most failures in the supraglottic larynx were man- ifest within 2 years, and all within 3 years (Fig. 1), apparent recurrences of glottic cancer continued to occur for 7 10 years, frequently on the contralateral cord, strongly suggesting that these late recurrences were indeed second primaries.

Discussion

Cancers of the head and neck region provide good illustrations of many of the potential causes of fail- ure of radiotherapy. Awareness of the various fac-

TABLE V

Cumulative appearance of failures for borderline histology lesions of the vocal cords compared with those for invasive squamous cell carcinomas of the vocal cords and supraglottic larynx 1952 through 1973 (analysis, July 1975).

No. failures/ No. patients

Time after treatment

I yr 2 yrs 3 yrs 5 yrs 7 yrs 10 yrs

Tl + T2 Vocal cords 0% 25% 58% 83 % 100%

(in situ) 12/86 (0/12) (3/12) (7/12) (10/12) (12/12)

Tt + T2 Vocal cords 41.5% 66% 74% 89% 94% I00%

(invasive carcinoma) 65/330 (27/65) (43/65) (48/65) (58/65) (61/65) (65/65)

T1 + T2 Supraglottic larynx 66.5% 93% 100%

(squamous cell) 15/96 (10/15) (14/15) (15/15)

From Pene and Fletcher 118].

62

tors tha t m a y con t r ibu te to t r e a t m e n t fa i lure is es-

sent ial when one p lans new the rapeu t i c s trategies.

F o r too long, r ad io the rap i s t s have been so obsessed

wi th the oxygen effect tha t o the r fac tors o f equa l or

g rea te r i m p o r t a n c e have been ignored . This is no t

to say tha t oxygen is u n i m p o r t a n t , bu t it mus t be

pu t in to perspect ive.

M a n y o f the po ten t i a l causes o f fa i lure are easi ly

r emed iab le wi th exist ing concepts and technology .

F o r example , i nadve r t en t u n d e r d o s a g e and geo-

g raph ic misses shou ld rare ly occur i f r a d i o t h e r a p -

ists are a ler t to the c i rcumstances when u n d e r d o s -

age is a risk, and a d e q u a t e qua l i ty con t ro l is ma in -

ta ined. Similar ly , concepts such as t i t r a t ion o f ra-

d i a t i on dose to the vo lume o f cancer present , avo id -

ance o f spli t courses tha t u n d u l y p r o t r a c t the du r -

a t i on o f t r e a t m e n t or unnecessa ry de lay be tween

surgery and p o s t o p e r a t i v e i r r ad ia t ion , and the use

o f c o n c o m i t a n t boos t s when there is evidence o f

r a p i d t u m o r g rowth are s imple to u n d e r s t a n d and

easy to a p p l y in cl inical pract ice .

On the o the r hand , there are some causes o f fail-

ure, e,g. the d e v e l o p m e n t o f second p r imar ies , and

recurrences due to the r a n d o m n e s s o f r ad i a t i on cell

ki l l ing tha t c a n n o t be ove rcome by any modi f ica-

t ion o f r a d i o t h e r a p e u t i c technique. There remains ,

however , a s ignif icant p r o p o r t i o n o f t u m o r s tha t

c a n n o t be con t ro l l ed with o p t i m a l exis t ing tech-

niques, in which the cause o f fa i lure is po ten t i a l ly

remediab le . R a d i o t h e r a p i s t s have a var ie ty o f new

s t ra tegies in thei r a r m a m e n t a r i u m , bu t up till now,

these have genera l ly been tes ted in an unselect ive

way on all " r e s i s t an t " tumors . W h e n a new s t ra tegy

is a imed at only one or two po ten t i a l causes o f fail-

ure, it is obv ious tha t the chances o f d e m o n s t r a t i n g

a benefit for the subset o f pa t ien ts who migh t be

he lped are d i lu ted by the inc lus ion o f pa t ien t s

whose cause o f fa i lure is un re l a t ed to the r a t iona le

o f the new t r ea tmen t s t ra tegy. The hope and chal-

lenge for the fu ture is tha t be t te r means o f ident i -

fying and m o n i t o r i n g po ten t i a l causes o f t r e a tmen t

fa i lure will be deve loped so tha t eventua l ly the ther-

apeu t ic reg imen r e c o m m e n d e d for a given pa t i en t

will be de t e rmined by the charac ter i s t ics o f his pa r -

t icu lar tumor .

Acknowledgment

This inves t iga t ion was s u p p o r t e d in pa r t by gran ts

CA06294 and CA16672 a w a r d e d by the N a t i o n a l

Cance r Ins t i tu te , D e p a r t m e n t o f H e a l t h and H u m a n Services.

References

1 Alper, R. (Editor) Cell Survival after Low Doses of Radia- tion: Theoretical and Clinical Implications. The Institute of Physics, John Wiley & Sons, London, 1975.

2 Barker, J. L. and Fletcher, G.H. Time, dose, and tumor volume relationships in megavoltage irradiation of squa- mous cell caricnomas of the retromolar trigone and anterior tonsillar pillar. Int. J. Radiat. Oncol. Biol. Phys. 2:407 414, 1977.

3 Barkley, H. T. and Fletcher, G.H. The significance of re- sidual disease after external irradiation of squamous cell car- cinoma of the oropharynx. Radiology 124:493 495, 1977.

4 Bataini, J. P., Bernier, J., Brugere, J. et al. Approche ra- diotherapique des adenopathies cervicales secondaires aux cancers epidermiques du larynx et du pharynx. In: Adeno- pathies Cervicales Malignes. Editors: J. Pinel and J. Ler- oux-Robert. Masson Publishers, Paris, 1982, 129 135.

5 Battermann, J. J., Breur, K., Hart, G. A. and Van Peperzeel, H.A. Observations on pulmonary metastases in patients after single doses and multiple fractions of fast neutrons and 6~ gamma rays. Eur. J. Cancer 17:539 548, 1981.

6 Charbit, A., Malaise, P. and Tubiana, M. Relation between the pathological nature and the growth rate of human tu- mors. Eur. J. Cancer 7: 307, 1971.

7 Cox, J. D., Byhardt, R. W., Komaki, R. and Greenberg, M. Reduced fractionation and the potential of hypoxic cell sensitizers in irradiation of malignant epithelial tumors. Int. J. Radiat. Oncol. Biol. Phys. 6: 37-40, 1980.

8 Dische, S. Hyperbaric oxygen: the Medical Research Council Trials and their clinical significance. Br. J. Radiol. 51:888 894, 1979.

9 Fertil, B. and Malaise, E-D. Inherent cellular radiosensi- tivity as a basic concept for human tumor radiotherapy. Int. J. Radiat. Oncol. Biol. Phys. 7:621 629, 1981.

10 Fields, R. S. and Hogstrom, K.R. A computer model for combining electron and photon beams. In: Proceedings of the Symposium on Electron Dosimetry and Arc Therapy, pp. 161 180. Editor: B. Paliwal. American Institute of Phys- ics, Inc., New York, 1981.

11 Fletcher, G. H. Clinical dose-response curves of human malignant epithelial tumours. Br. J. Radiol. 46:1 12, 1973.

12 Fletcher, G. H. Combination of irradiation and surgery. In: International Advances in Surgical Oncology, 2:55 98. Editor: G. P. Murphy. Alan R. Liss, Inc., New York, 1979.

13 Fletcher, G . H . Textbook of Radiotherapy, 3rd edn. Lea E Febiger, Philadelphia, 1980.

14 Fletcher, G . H . The scientific basis of the present and fu- ture practice of clinical radiotherapy. Int. J. Radiat. Oncol. Biol. Phys. 9, in press, 1983.

15 Henk, J. M. and James, K .W. Comparative trial of large and small fractions in the radiotherapy of head and neck cancer. Clin. Radiol. 29:611 616, 1978.

16 Parsons, J. T., Bova, F. J. and Million, R .R . A re-evalu- ation of split-course technique for squamous cell carcinoma of the head and neck. Int. J. Radiat. Oncol. Biol. Phys. 6: 1645 1652, 1980.

17 Paterson, R. The Treatment of Malignant Disease by Rad- ium and X-rays. Williams & Wilkins Co., Baltimore, 1949.

18 Pene, F. and Fletcher, G .H . Results in irradiation of the in situ carcinomas of the vocal cords. Cancer 37:2586 2590, 1976.

19 Peters, L. J., Withers, H. R., Thames, Jr., H. D. and Fletcher, G . H . Keynote address: The Problem: tumor radioresist- ance in clinical radiotherapy. Int. J. Radiat. Oncol. Biol. Phys. 8:101 108, 1982.

20 Sinclair, W. K. and Morton, R.A. X-ray sensitivity during the cell generation cycle of cultured Chinese hamster cells. Radiat. Res. 29:450 474, 1966.

63

21 Thames, H. D., Peters, L. J., Spanos, W. and Fletcher, G. H. Dose response of squamous cell carcinomas of the up- per respiratory and digestive tracts. Br. J. Cancer. 41, Suppl. 4: 35-38, 1980.

22 Thames, Jr., H. D., Peters, L. J., Withers, H. and Fletcher, G .H. Accelerated fractionation vs hypei'fractionation: ra- tionales for several treatments per day. Int. J. Radiat. Oncol. Biol. Phys. 9: 127-138, 1983.

23 Vikram, B., Strong, E. W , Shah, L and Spiro, R. H. Elective postoperative radiation therapy in stages III and IV epidermoid carcinoma of the head and neck. Am. J. Surg. 140:580 584, 1980.

24 Weichselbaum, R. R. and Little, J .B. The differential re- sponse of human tumors to fractionated irradiation may be due to a post-radiation repair process. Br. J. Cancer 46: 532 537, 1982.

25 Withers, H. R. and Peters, L .J . The application of RBE values to clinical trials of high LET radiation. In: High LET Radiation in Clinical Radiotherapy, pp. 257 261. (A supple- ment to Eur. J. Cancer), 1979.

26 Withers, H. R. and Peters, L.J. Biologic aspects of radia- tion therapy, In: Textbook of Radiotherapy, 3rd edn., pp. 103 180. Editor: G. H. Fletcher. Lea & Febiger, Philadel- phia, 1980.