ClinicalRadiology(1980) 31, 121-135
© 1980 Royaltxouegeof Radiologists
,End-points in Cancer Therapy' The Glyn Evans Memorial Lecture of the Royal College of Radiologists* ARTHUR JONES
Department of Radiotherapy, St Bartholomew's Hospital and Medical College, London
Mr President, Ladies and Gentleman, Clinicians would I think be reluctant to accept, with Helmholtz, that 'all science is measurement': nevertheless the remarkable advances in medicine of the last 30 years have depended in great measure on quantitation. While the individual disciplines which make up oncology can be as precise in methodology as any others in clinical science, the ultimate assessments of therapy have in the nature of things often been less substantial. The most sophisticated of clinical trials can only be as good as its data, its endpoints and the perspective of their interpretation. A look over the nature and diversity of some of the endpoints adopted by different disciplines may therefore be timely. I would like to thank you, Mr President, and the Council of the Royal College of Radiologists most sincerely for the honour of being invited to give this Glyn Evans Memorial Lecture, and for the privilege of discussing this topic before a multidisciplinary audience at Cambridge. May I also take this opportunity of thanking my friends in various branches of oncology who by their stimulating discussions have made this review possible? The subject would I believe have appealed to Glyn Evans, who by his vision and achievement did so much to further the treatment of patients with cancer. He not only built up the outstanding Radiotherapy Centre at Velindre Hospital in Cardiff, but in a wider sphere he was the first medical adviser of Tenovus, which has had such a beneficial influence on cancer research in many centres. Glyn Evans was essentially foward4ooking, and he would be showing his delight at the progress which is now being achieved on such a broad front. Survival statistics in cancer are under discussion, not to say under fire, from both epidemiologists and clinicians, who are in fact each looking at extremes of a diversity of problems. In the community at large mortality rates and incidence rates of cancer are highly correlated, while neither is related to the actual survival rates. Engstrom and Austin (1977) have pointed out that survival rates should not be *Deliveredat Cambridge, 16 September 1978. 10
used as the sole or primary measurement of progress in national cancer control because factors unrelated to the efficacy of treatment play an important part in determining those rates and their trends. A telling instance is the improved survival reported by clinicians for most forms of leukaemia, while the global survival for all forms of leukaemia may in fact have remained constant because of increasing incidence of the highly fatal forms; and the converse situation seems to hold for carcinoma of the prostate. From the clinical point of view, for the individual 'tumour species' of defined type and stage the timehonoured survival rate, say at five years, provides very incomplete information even when 'corrected' for age and so on; for it fails to distinguish between early and late deaths, and it wastes information on patients followed for less than the whole period. There is no doubt that a 'life table' is the most accurate description of a set of data on the times to death of a group of patients. This gives of course an estimate of the proportion who will still be alive at different times after entry, calculated with due allowance for incomplete follow-up: essentially it characterises the survival pattern of the series. It requires regular determination of the status of all patients at risk, and it permits use of all data; its major disadvantages are (a) the requirement of a large study sample in order to provide reasonable confidence limits as the analysis proceeds and the number of individuals at risk diminishes, and (b) that patients may also die from disease not directly related to neoplasia or from causes induced by therapy. Nevertheless, correctly used and interpreted, the Kaplan-Meier method is the most useful available for series characterisation, and two points of difficulty related to misuse and misapprehension have therefore to be mentioned. As the method uses probability theory and the term 'projected' or 'expected' five-year disease-free survival is employed, this engenders scepticism in the minds of clinicians as to how many patients had actually survived for the five years; and this actual versus actuarial controversy demands that the survival by the 'direct method', presenting the results at the end of a period, should also be given at the same time.
The second difficulty arises from its use in a small series and too early in the course of a disease, thereby spuriously bolstering the results of a new method of treatment. A similar situation may follow publication of a too-early analysis of a major trial, to show a temporary difference which is later smoothed out by larger numbers, but which in the interim may have an unconscious bearing on practice outside the trial. That this can occur also with the direct method is exemplified in the data of the excellently conducted CRC breast study which compares survival and recurrence after mastectomy, both with and without postoperative X-ray therapy. Murray (1974), giving a preliminary report in the British Journal o f Surgery, was careful to point out that at four years too few patients had been followed for too short a time to give meaningful results; but he also mentioned that the divergence of the graphs showed a difference approaching statistical significance, suggesting 'that the occurrence of distant metastases is higher in the group treated with irradiation'. Of course, with time the apparent difference disappeared and the recent report (Leiberman, 1978) showed no difference in survival but a useful reduction in local recurrence in those irradiated. Changes in diagnostic criteria and in the detail of delineation and characterisation, as well as in supportive medical care, have been shown to alter significantly the apparent natural history of many types of neoplastic disease. For instance, in Hodgkin's disease lymphopenia and immunological anergy at presentation were formerly regarded as bad prognostic features, but have lost this significance in modern staging; and an initial thrombocytopenia in acute myelogenous leukaemia has been robbed of its former dire import by the advent of platelet transfusion. So that despite the arguments put forward by Gehan and Freireich (1974) for the use of historical controls, once a pilot study has been completed, and unless large differences in responses are to be expected, there is rarely any alternative to a well-designed randomised clinical trial; and the publication last year of the MRC Report by Peto et aL (1976, 1977) is a landmark in the analysis of such trials.
is similar to that of a normal population of the same sex and age constitution'. This concept Was elaborated by Easson and Russell (1963) and by Easson (1966) particularly in relation to Hodgkins disease; and it has had such an influence on oneo. logical thought that by 1977 Carter (1977), a medical oncologist, wrote 'by "cure" is meant that the life expectancy of the treated cancer patient is the same as the "normal" life expectancy - specifically the same as that in a matched cohort of the general population'. The concept has been valuable in three respects: in focusing attention on the actual possibility (and in Hodgkin's disease now the probability) of cure; in reassuring patients; in the public relations of cancer research; and possibly in another particular which I shall mention at the end of this lecture. However, it is now sufficiently venerable to withstand further examination. At different points along the time scale it has been applied for instance in studies of lymphoma, uterine carcinoma and breast carcinoma Bush et al. (1977) in Toronto (Fig. 1) analysed a large series of patients with non-Hodgkin's lymphoma, to define those who could be cured by local irradiation, and found that Stage IA patients had a survival curve compatible with cure after six years (although he conceded that a 15- or 20-year follow-up would be a better criterioh). For carcinoma of the cervix uteri (Fig. 2), randomised between cobalt-60 or 22MeV Betatron irradiation, again at Toronto, Bush (1979) compared each series with the slope of a normal survival curve: the rate of dying followed the normal curve after about six years for Stage lib, and after about eight years for Stage III. In Brinldey and Haybitfle's classical paper (Fig. 3) on the long-term follow-up of patients with breast cancer in Cambridge I00
e ~ o ~
• l-\~a, " ~ | ~ .
N°rmal I ~
\ " . ' ,~. ~O
"% . . . . . . . . I ~
CONCEPTS OF CURE The idea of 'cure' in cancer was for generations linked to arbitrary and f'mite survival periods, often 5 or 10 years, but some 20 years ago Russell (1958) of Manchester, modifying an idea of Greenwood (1926), enunciated a new definition: 'We can speak of cure when in time - probably a decade or so after treatment - there remains a group of disease-free survivors whose progressive death rate from all causes
Yr a f t e r Dx
Fig. 1 - Non-Hodgkin's lymphoma (Toronto series) actuarial survival by stage. (Reproduced from Bush et al (1977) Cancer Treatment Reports, by kind permission.)
E N D - P O I N T S IN C A N C E R T H E R A P Y
increase mortality. As pointed out by Frei and Gehan (1971), if 50% of lung cancer patients could be 'cured' of their tumour, they would still have (5o, excess mortality from smoking-related diseases. But 50 the third difficulty is conceptual, in that such recorded 40 ~'°"e-"e"°-'--o-,..t._. co6° (52) achievement depends inversely on progress in what30 ever factors determine the longevity of the whole population. Easson (1966) compared his series of ~- zo Hodgkin's disease with an 'expected' survival curve prepared from the life tables appropriate to the northSlope o f n o r m Q I surVWQI c u r v e west region of England; Brinkley and Haybittle June ~ l I I I I / I (1975) compared their breast carcinoma series with I0 2 4 6 8 " I0 12 14 the lower than average death rate of East Anglia; Yrs efter Dx Fig. 2 - Carcinoma of cervix uteri (Toronto series) - actuarial while Bush used the Ontario figures from Statistics survival of all patients in Stage IIB randomised between Canada. Absolute international criteria are thereby cobalt-60 and Betatron irradiation. (Reproduced from Bush, difficult to establish, and base4ines may change more R. S. (1979) Malignancies of the ovary, uterus and cervix. rapidly in certain parts of the world than in others. In The Management of Malignant Disease Series, ed Peckham, M J. & Carter, R. L. Edward Arnold, London, by kind Incidentally, it has been suggested that as very longterm survivors from breast carcinoma have the same permission.) life expectation as the 'normal' population, but still 16 times the risk of death from breast carcinoma, (Brinkley and Haybittle, 1975), after 21 years the they could be less at risk of dying from other causes. survival curve of the breast cancer patients paralleled Could this be due to indirect benefits of screening the expected survival curve of a normal population. by continued follow-up observation rather than to Extrapolation indicated that the proportion 'cured' enhanced immunity? was about 18% of all patients and about 30% of It would appear that the 'Greenwood-Russell clinically early cases. principle' is of most value in describing diffuse neoThere are three difficulties about the 'Greenwoodplastic processes or those with marked metastatic Russell principle'. The obvious one is that patients potential, lacking an early 'cut-off' in the survival in the 'cured' group may still die of their disease. curves, and in which late toxicity of treatment may Brinkley and Haybittle found that after 20 years be a significant factor in determining the length of there was a 16-fold excess of deaths from breast survival. The more effective the treatment, the less cancer compared with the normal population;and late the need to apply this concept, and even in Hodgkin's manifestations of Hodgkin's disease are also well recog- disease through which it was developed it is now no nised especially in the inadequately staged disease. longer used. This engenders scepticism in other specialities on the The rapid progress of therapeutic research has probity of oncological methods. Secondly, excess brought calls for methods of determining the probadeaths do not necessarily result from the original bility of cure as early as possible, for cancer chemotumour, if this is associated with other factors which therapists do not like to have their patients 'locked up' in one trial for too long. In experimental in vivo tumour systems Skipper et al. (1964) showed that the ~pu)atlon ef period from end of treatment to death was a precise SURVIVAL measure of the number of neoplastic cells persisting at the end of treatment. Furthermore in those homogeneous transplanted tumour systems, survival beyond a given point after treatment was associated almost invariably with cure (Frei and Gehan, 1971). For certain human cancers, patients who remain relapse-free for a relatively short time after treatment can be regarded as cured: varying (with 95% confiY~AR5 dence limits) from two years for Wilm's turnout to Fig. - Breast carcinoma (Cambridge series) - survival rate five years for localised endometrial cancer. These afte treatment in 704 cases compared with expected survival three principles underlie the medical oncologists' of normal population of same age distribution. (Reproduced emphasis on the 'relapse-free interval' in relation to from Brmkley and Haybittle (1975), Lancet, by kind permission.) concepts of cure. 100 9O 8O 7O 6O ; . . . . .
e L Q
. . . . . . . . . . . . .
age dlsttlbUt on
plotted against radiation dose, has been a powerful experimental tool and has now been extended (Salmon, 1973; Salmon etal., 1978) in human cancer chemotherapy as a predictive test of different sensi. tivity. The in vivo assay systems such as the lung colony system of Hill and Bush (1969) are similarly applicable to investigation of radiation or drugs. The most important of the experimental methods avail. able in mammalian radiobiology are: animal survival, in situ tumour measurements, and direct measure. ments of cell survival in vitro and in vivo (after excision of irradiated tumour) (Table 1).
zoo 4bo 6~o s~o i~oo i;oo f~oo t&o i~oo z~oo z~oo
Table 1 - Experimental methods for measuring radiation effects on tumour growth
Animal survival Effects on turnout growth (a) Local control series). Kaplan-Meier product limit estimates of probability (b) Regressiontime to ½ of being in remission, circles indicating duration of remission. (c) Regrowth delay (Reproduced from E. DonnaU Thomas et al. (1977) LeuDirect measurement of cell survival kaemia-Research, by kind permission.) In vitro and in vivo after excision of irradiated tumour Under the title of 'cure of leukaemia by marrow transplantation' Thomas and his colleagues (1977) In situ measurements exploit the pattern of from Seattle discussed their experience of 110 regression, growth delay and recurrence, and while patients with acute lymphoblastic or myelogenous regression rate is characteristic of the biological leukaemia who were given marrow from HLA- make-up of the tumour Thomlinson (1977) found identical siblings after high-dose cyclophosphamide regrowth measurements to be more significant in and 'supralethal' total body irradiation (Fig. 4). The elucidating the pattern of response. Clearly the Kaplan-Meier actuarial analysis of the survival curve important question is how these methods correlate of these patients showed three phases, the last of with each other. Barendsen and Broerse (1969), which extending from two to 6½ years was almost using a rat rhabdomyosarcoma, compared volume flat, with negligible loss of patients and no recurrent responses and cell proliferation simultaneously and leukaemia. This flat portion of the curve (about found a remarkable agreement between the proper. 15% of the patients) constituted their operational tion of tumours cured and the extrapolation of the definition of cure in these patients: in this un- survival curves. On the other hand, McNaUy (1975) controlled study 'cure' is used in a restricted sense. using a rat fibrosarcoma showed that the modifying Further observation over a few years will in acute effects of oxygen and radiation quality on tumour leukaemia test the validity of this thesis, but it will growth delay and on cell survival in vitro (after take many years to assess the absence of long-term irradiation in vivo) were different. More recently toxicity which is also an essential component of (McNally and Sheldon, 1977) he has shown that for ~ c u r e ~. a murine tumour neither the cell survival assay nor the growth delay agreed with the TCDso (dose to TUMOUR RESPONSE RELATIONSHIPS AT cure 50% of the animals) when investigating the EXPERIMENTAL LEVELS effect of the hypoxic cell sensitiser misonidazole. As things may seem simpler in test-tubes or in Clearly there is considerable variation from one mice it may be of interest to see how the experi- animal tumour to another, let alone any extramentalists actually fare with their end-points. The polation to man, and Steel (1977) has highlighted as important tumour cells - the stem or clonogenic a deficiency in our knowledge the relationship cells, which are capable of sustained replication -- between clonogenic cell survival and the gross endoften make up less than 1% of the cells of an experi- points of growth delay and tumour cure. In fact, the mental solid tumour, yet it is the behaviour and fate 'Quantitation of Tumour Response' has been selected of these key cells which have to be inferred in the as the theme of the next L. H. Gray Conference to be number of experimental methods which are now held here in a year's time, so clinicians may obtain available. The in vitro colony assay (Puck and Marcus, some solace from knowing that the subject is far from 1956), in which loss of reproductive integrity is being tied up, even at the laboratory level. Fig. 4 - Acute leukaemia treated by marrow transplant after chemotherapy and supralethal total body irradiation (Seattle
TIlE CLINICAL EVALUATION OF TUMOUR RESPONSE The objective evaluation of tumour response in an i~dividual patient depends of course on clinical observations, imaging techniques, and biochemical determinations, and these must be viewed in context to obtain a clinically meaningful appraisal. Apart from the very special instances of intra-ocular tumours, those of the skin (both primary and metastatic) provide opportunities for the greatest accuracy of physical measurement, as shown by the elegance of Thomlinson's studies (Thomlinson, 1977)based on r~any thousands of caliper readings; and even these reveal that what is being measured is usually the clinical tip of a pathological iceberg. Squamous carcinoma of the upper respiratory tract is also accessible, and when turnouts are small a fairly high degree of accuracy is possible in their measurement. Once the carcinomas have become large and diffusely infiltrating there is however little precision in the assessment, and it is well to note that many studies of chemotherapy are of necessity based on such material. If responses of accessible lymph node masses are difficult to quantitate, how much more difficult is the situation with abdominal and pelvic nodes. When we clinically observe ovarian cancer and infiltrating carcinoma of the colon and rectum the accuracy of physical measurement is often least, and these again are two important diseases which are the subject of many clinical trials. In this context, Moertel of the Mayo Clinic (Moertel and Hanley, 1976) has reported on the effect of measuring error on the results of therapeutic trials in advanced cancer. Sixteen experienced oncologists each measured 12 simulated tumour masses (some of which were identical in size); the lesions varied from about 2 to 15 cm in diameter and were covered with foam rubber ½ - 1 ½ i n thick to mimic subcutaneous tissue or the abdominal wall. The oncologists were asked to measure each sphere using the technique usually used in practice. They found the 'placebo response rate' in unchanged masses to be 25% if a 25% regression was employed, and 7% if a 50% criterion was taken. Moertel points out that when an oncologist is measuring a deep-seated irregular mass through the muscle guarding of an uncomfortable patient, the error of his measurement will undoubtedly be far greater than that recorded in this study. 'Certainly under real life circumstances a 25% regression must be considered meaningless, and even a 50% regression is questionable.' Such critical approaches to determination of tumour size should not of course be construed as observational nihilism, for it is important that clinical observations should be pursued to the limit of
accuracy of that particular method in the individual patient - providing that the observations are relevant and reproducible. These may often be clinical observations of the effects of the turnout, direct or indirect, rather than of the tumour contour itself. For instance the neurological effects of primary or metastatic tumours may be documented and quantitated with some accuracy. Even the relief of a compressive effect of bronchial carcinoma on the pulmonary artery can be documented by phonocardiography (Fig. 5)with loss of the systolic murmur. Hormonal effects of tumours may be readily quantified biochemically, but comparative changes at the clinical level are less easily documented. In acromegaly the regression of soft tissue (following Bromocriptine and pituitary irradiation) may now be recorded by three-dimensional contour scanning (Fig. 6). And in Cushing's disease due to adrenal carcinoma the response to Cyproterone acetate may be quantitated in terms of the response of hair growth (Burgess and Edwards, 1978). Of considerable practical importance is the control of serous effusions, and changes in the frequency and BEFORE DXR ECG (heartrate 100) ~
!L&tL'l"ri/I,~ l JlrhJJr'z, . . . . . I
,I,,. ~ L
r tr~,g ,L..
.i~r~* L L',
,t r ii '
-J~ EcG I
CarottdPulse DXR RESPONSIVE PULMONARY
Fig. 5 - Carcinoma of the left main bronchus with systolic murmur of functional pulmonary stenosis. Disappearance of murmur documented by phonocardiographybefore and after X-ray therapy (4000 rad/25f/35d).
FOR THREE MONTHS
Fig. 6 Acromegaly: response to bromocriptine documented by stereoscopic contour mapping.
volume of aspiration are often used as evidence of therapeutic efficacy. This may be so in the thorax (Fig. 7); but while ultrasound can readily reveal ascites the indications for paracentesis abdominis are so variable, and readjustment of visceral hydrodynamics so readily occurs, that only gross alterations in frequency of aspiration can be regarded as significant. In view of the limitations of purely physical measurement it is not surprising that clinicians should turn avidly to imaging techniques, the most important of which have been radiographic. Our knowledge of the doubling times of many human tumours and their modification by therapy is essentially derived from work on the measured radiographic changes in pulmonary metastases, the accuracy of which is facilitated by both their spherical growth and their demarcating contrasting density. Objective evaluation in the common primary lung cancer is more difficult, for at least half of inoperable cases have disease which is technically 'non-measurable' by calipers (Dawson et al., 1966), but even these have been used in an investigation with the unlabelled radiographs placed in ranking order by different observers. Apart from the study of pulmonary metastases most of our radiologically derived knowledge of human tumour growth rates and response comes from lymphography. Its accuracy in diagnosis is well known, and while the contrast remains for about 18 months and can if necessary be refilled, the value of the method in assessing the retroperitoneal mass is limited, for both in the lymphomata and in testicular tumour metastasis it often significantly underestimates the volume when compared with CAT scanning (Schaner et al., 1977). Measurements of regression of moderately advanced disease based
k (a) RE LATIONSHIP BETWEEN MEDIASTINAL IRRADIATION & ASPIRATION OF LEFT CHYLOTHORAX MEDIASTINAL XRT
VOLUME ASPIRATE LITRES
(b) Fig. 7 -
Mediastinal lymphoma with left chylothorax.
Response to mediastmal irradiation shown in frequency and volume of pleural aspirate.
solely on lymphograms are liable to be similarly skewed. Increasingly the specialised delineation and serial observation of tumours is by computer assisted tomography, which in the brain and orbit has produced such reliable results. While its accuracy in the thorax and abdomen (Fig. 8) is also high, .and it can provide repeated observation of previously inaccessible sites, it is at present hampered by its inability to detect tumour deposits in normal sized or minimally enlarged lymph nodes. Nevertheless its delineation of pelvic tumours has given a new order of accuracy to their assessment. Although discrete masses may be seen to regress it is less usual for an 'infiltrative'
IN C A N C E R
Fig. 9 - Parieto-occipital metastasis of anaplastic bronchial carcinoma: response to whole-brain irradiation shown by technetium brain scans, before and two months after treatment. Fig. 8 - Response of Hodgkir~'s disease in para-aortic lymph nodes to MVPP chemotherapy, documented by CAT scannmg before and three months after treatment.
appearance to revert to normal, rendering assessment of efficacy of treatment also less certain. In the formidable array of newer forms of imaging techniques the value of serial quantitative observations varies, because of the innate physical limitations of each method, and this must be remembered when assessing the same neoplastic disease as it affects different systems and in comparing the end results. Radionucleide investigations have the limitations of signal : noise ratio of the radiopharmaceutical used. This explains why 3 mm bone lesions can be seen on a bone scan but why 2 cm lesions may be missed on a brain scan (Ell et al., 1978). So that a negative brain scan (Fig. 9) after cranial irradiation for cerebral .metastases, while useful as collateral evidence of tmprovement, has no significance for absolute remission. In the common use of bone scanning with
tectmetium-99m phosphate for breast carcinoma metastases, improvement or deterioration is readily observed but difficult to quantitate unless the scan becomes normal. (The value of serial bone scans in prostatic carcinoma has been shown by Shearer et al. (1974) who found that those whose bone scans became negative after at least three months on hormone therapy were more likely to survive at least two years than those whose bone scans remained positive: and the failure of an abnormal scan to revert to normal could be the first indication of hormonal resistance.) In liver scanning with technetium-99m sulphur colloid for metastases (Fig. 10) individual focal abnormalities may be difficult to recognise at repeated scans, and lesions of less than 0.5cm in diameter are not usually detected; but within these limitations it can provide useful documentation of therapeutic response and relapse, if not a quantifiable form.
Fig. 10 - Hepatic metastasis of breast carcinoma: initial response to adriamycin and cyclophosphamide, and later relapse documented by liver scans using technetium antimony sulphide colloid.
BIOCHEMICAL MONITORING AND TUMOUR MARKERS Human tumours release a wide variety of products which, as well as accounting for some clinical features, provide means of monitoring disease activity and the effect of treatment (Neville and Cooper, 1978). Gonadotrophins have long been measured for the monitoring of tumours with trophoblastic elements (Bagshawe, 1969, 1976) and these, and the homogeneous irnmunoglobulin synthesis of plasma cell neoplasms (Salmon, 1973; Alexanian, 1977) are the models par excellence for this approach. Suitable biochemical markers fall into two groups, first the products of tumours which may appear in plasma or urine, and secondly the tumour-associated changes which may involve a variety of effects particularly on enzymes and proteins. Of the tumour products some are oncofoetal and few are tumour-specific. While carcinoembryonic antigen (CEA) has been used to detect recurrence or metastasis in carcinoma of the colon or urothelium (to give a 'lead time' on the clinical signs), when used to monitor chemotherapy the decline in CEA level does not reflect quantitative changes in tumour volume in the way in which HCG assays mirror alterations in the number of cells of gestational choriocarcinoma (Bagshawe, 1969). Alphafetoprotein (AFP), once thought to be specific for hepatocellular carcinoma, is produced by a variety of tumours, and while almost all hepatomas produce AFP the plasma levels seem unrelated to tumour size, the highest being associated with nodular and poorly differentiated tumours. AFP assays have nevertheless a useful role also in gonadal, gastric and pancreatic turnouts. In teratomas, successful treatment (by surgery, irradiation or chemotherapy) can result in an AFP fall to normal; but in
gastrointestinal cancer recurrences can develop with. out rising AFP titres. The practical conclusion is the importance of assaying more than one parameter, for instance CEA and AFP for gastric carcinoma, and AFP and HCG for teratomas. Of the hormones, HCG and HCG-like molecules are produced by a variety of tumours, and the levels in plasma and in CSF can be used to monitor the treatment of pineal tumour as shown in Fig. 11. Of the tumour-derived products, serum acid phosphatase levels are elevated in 90% of prostatic carcinoma with bone metastasis, and this is a useful monitor of therapy in the short term; but later on, extensive metastasis can develop under treatment without any associated rise in acid phosphatase activity. While in any advanced metastatic liver in. volvement there are gross changes in GGT (gammaglutamyl transpeptidase) activity, raised alkaline phosphatase and transaminases, these tests are insensitive to small tumour loads in the liver. Finally, of the non-specific estimations, urinary hydroxyproline OHP, derived from collagen, twothirds of which is found in the skeleton. OHP excretion is increased with increased bone turnover, and Powles et al. (1975) have shown that the OHP/ creatinine ratio is of value in assaying the efficacy of chemotherapy and endocrine measures in controlling bone metastases. The problem is in transferring such a variety of biochemical data to the comparative clinical context. For the monoclonal myeloma, Salmon (1973) has used quantitative studies of M-component synthesis to calculate the total body burden of tumour cells. Apart from the innate interest of the 'marker kinetic' approach in oncology, the response to chemotherapy of myeloma (Alexanian, 1977)hasbeen usefully documented in terms of the pretreatment turnout mass;
END-POINTS IN CANCER THERAPY SERUM AND CSFHCG DURING RADIOTHERAPY
HC6 ~fnl O
CSF HCG ng/rnI
but for other tumours, with their multiple production o f ectopic factors and many reactive changes, multiparametric analysis is required for the biochemistry, and the complexities of the necessary statistical ° analysis o f time-based longitudinal studies are still in their early days o f development (Neville and Cooper, 1978). What is quite clear from the look at clinical, imaging and biochemical methods of estimating tumour response is that no single criterion can be used in all neoplastic diseases, nor even at all stages in the same disease. With so many relevant parameters 5000, measurable, it is the pattern of change that matters, and our efforts in future need to be directed to the recognition and quantitation of changes in variegated patterns. RESPONSE, REGRESSION AND REMISSION The clinical effects by which responses to treatment are recorded have three components: in space, with changes in measurable lesions (and in those 'evaluable' but not measurable); in disturbance of function, noted clinically and by biochemical changes; and in time, in the dates of assessment and in both the duration of the response and the 'time to recurrence'. The work of Moertel on clinical observer error has already influenced the criteria for describing measurable and reproducible decrease in lesion size in a specified period of time, and this is incorporated in the criteria set out by Carter, (1977) in his thoughtful review o f approaches to clinical trials (Table 2). Similar criteria have been adopted in assessment of advanced breast cancer by the UICC project programme (Hayward et al., 1977) and this with its incorporation of base-line studies, detailed recordings and definitions o f response, should serve as a model for similar cooperative studies in the future. Table 2 - Criteria for objective response and disease progression in solid tumours (from Carter, 1977)
(b) Fig. 11 - Pineal germinoma producing HCG: progress during and following cranial irradiation monitored both by estimation of HCG in blood (a) and in CSF, and (b) by CAT Scanning.
Complete disappearance of all demonstrable disease
I> 50% reduction in the sum of the products of the longest perpendicular diameters of discrete measurable disease, with no demonstrable disease progression elsewhere
No change in the size of any measurable lesion or < 50% reduction of measurable disease as defined above
> 50% increase in the sum of the products of the largest perpendicular diameter of any measurable lesion
There is clearly such a need, for in practice there have been vast differences in usage between the various subspecialties of ontology, so that a cursory look at published results may mislead those not involved in that particular type of study. For instance, the EORTC Clinical Screening Group (1973), reporting the 'oncostatic effect' of epipodophyllotoxin VP16213 in leukaemia and solid tumours defined remission as the 'apparent disappearance of the disease', regression as the 'apparent complete disappearance of tumour'; both remission and regression could be complete or incomplete, and whether each was more or tess than 50% was 'necessarily subjective'. Again, the term 'remission' has come to have specific import in the leukaemias and in lymphoma, but it has been extended less substantially to, for instance, myeloma when Salmon (Alberts et al., 1976) defines 'complete remission' as greater than 75% reduction in tumour cell mass, with relief of pain and increase in haemoglobin and serum albumen. In Hodgkin's disease patients are considered to be in complete clinical remission (Sutcliffe et al., 1978a) when there are no symptoms or signs of the disease, chest X-ray and lymphography are normal, and routine blood tests have become normal. The prognostic significance of such remission at one year is highly correlated with survival, and it has important implications for continuing chemotherapy. So it is of particular interest that in the recent investigation by Sutcliffe et al. (1978b) of post-treatment taparotomies after chemotherapy, they showed that 4 to 19 cases, considered to be free of abdominal disease on the strict criteria given, had residual active Hodgkin's disease, and that they were able to define the three histological categories negative, active, and healed disease. In most otber diseases it is unusual for patients labelled 'in remission' to be subjected to such searching checks of disease status, but one may surmise on the possible implications. The common usage of 'remission' in medical oncology having been defined, it is perplexing to find it used in CNS tumours in a quite different sense. Rosenblum et al. (1973) in describing their use of CCNU in malignant brain tumours developed a staging system depending on both symptoms and signs, and 'if a patient improves in disease stage he was considered a "Remission" '. By this criterion their initial clinical response to CCNU was 'a 37% remission rate' but this is liable to misinterpretation when reported briefly and out of context. The concept of 'disease stasis' has had a vogue in describing the unmeasurable, particularly in abdominal tumours, but Wiltshaw and Kroner (1976) pertinently state that in their experience of ovarian
carcinoma so-called 'no change' usually represent~ advancing disease. The methods for assessing tt~s tumour are probably the least sophisticated in the oncological spectrum, and this is unfortunate in viev¢ of its interest for immunotherapy and particularly for the recently reported predictive tests for cancer chemotherapy (Salmon et al., 1978). With increased specialisation in oncology and the mass of reports it becomes increasingly difficult for physicians to critically evaluate each study, and Carter (1977) has stressed the need for guide-lines for study design and data reports 'to bring some e0. herence to the chaos now found in the oncological literature'. The first steps out of a semantic rnor~ are being taken by strict def'mition of terms, but it Is clearly important that those who use them should be aware of the usages of others who may have more than a marginal interest in their study. EFFECTS ON NORMAL TISSUES When we turn to the other side of the therapeutic equation, to assessing the effect on normal tissues,it is important that we compare like with like. The tumour process is assessed by its main function -that of unbridled reproduction - so that normal tissue effects should also be assessed in terms of function rather than on any static histological appearances This gives us a diverse series of end-points depending on bodily systems, but in addition to specialised function in the limit the important end-point is the ability to cope with stress. Examples are seen in the haemopoietic reserve after chemotherapy, or the ability of connective tissue to respond to physical or ischaemic stress after irradiation. The quantitati0n of normal tissue responses has of course to be by the methods of the 'system specialty', and as these become progressively more sophisticated it follows that the evidence for levels of tissue damage has to be re-examined every few years; but the end-point chosen should be directly relevant to dose-response, and not some trivial side effect seized upon by the organ specialist: for what can be measured will be measured. In the case of late effects it is clearly desirable to have exact knowledge of the original patient pool, and adequate numbers of patients surviving long enough for latency and 'at risk' factors to be elucidated. Then not only may frequency diagrams be produced, but 'probit analysis' can pro. duce curves predicting the probability of developing the late complication (e.g. Wara et al. (1975) for thoracic radiation myelopathy) and this is of course what the clinician really wants to know. The real difficulties arise in the quantitation of combined radiation and chemotherapy effects in
END-POINTS IN CANCER THERAPY critical tissues of the antibiotic agents which enhance radiation reaction, and of those drugs such as Adriaroycin arid Methotrexate which themselves damage the tissues (the heart or CNS) directly. These effects are particularly important in paediatric ontology. As the 'dose effect factor' (Phillips and Fu, 1976)varies from 1.1 to 1.8 there is considerable urgency in acquiring further information from experimental models and clinical data. Finally, with increased survival, the problem of late carcinogenesis, exemplified in our recent experience by such bizarre tumours as mesothelioma and fibrosarcoma of the uterine cervix following prolonged alkylating therapy for Hodgkin's disease. Again we need data on incidence and latency in relation to the degree of exposure, and then on the probability of tumour induction; now the urgency is of course the need for guidance in matters o f adjuvant chemotherapy for curable conditions. THE PATIENT AS A WHOLE I have talked o f tumours, I have talked o f normal tissues, but what o f the patient as a whole? The emphasis in clinical trials being on objectivity, the 'quality of life' has often been regarded as being too subjective for meaningful evaluation. But two modern developments make an evaluation of such 'quality' of survival essential. When advances in treatment have so altered the natural history of a condition that for a 10ng period it may be 'morbid rather than mortal', the 'quality' o f life during remission is clearly o f prime importance. We can visualise a situation (Fig. 12)
04 PROBABILITy OF SURVIVING
Fig. 12 - The quality of survival: hypothetical survival curves for two methods of treatment which alter the pattern of survival without preventing ultimate demise. The difference (shaded) in 'areas under the curve' may conceal vast differences in well-being or in suffering.
when a combination of modalities of therapy significantly alters the pattern of survival while merely postponing the inevitable demise; the difference in 'areas under the curves' may conceal vast differences in well-being or in suffering which need to be elucidated. Secondly, at the more sanguine end of the therapeutic range, the quality o f survival o f those actually cured is a prime concern of patients, o f families and of society. Attempts at quantitating the totality of function have been based mainly on those of Karnofsky (1948) (Table 3a) who introduced the concept of Performance Status - a measure of the patient's ability to
Table 3a - Performance status rating (after Karnofsky and Burchenal, 1948) Defmition
Ableto carry on normal activity and to work; no special care is needed
Normal; no complaints; no evidence of disease
90 80 Unable to work; able to live at home, care for most personal needs; a varying amount of assistance is needed
Unable to care for self; requtres equivalent of institutional or hospital care; disease may be progressing rapidly
50 40 30 20 10 0
Able to carry on normal activity; minor signs or symptoms of disease Normal activity with effort; some signs or symptoms of disease Cares for self; unable to carry on normal activity or to do active work Requires occasional assistance but is able to care for most of his needs Requires considerable assistance and frequent medical care Disabled;requires special care and assistance Severely disabled; hospitalisation is indicated although death not imminent Very sick; hospitalisation necessary, active supportive treatment is necessary Moribund; fatal processes progressing rapidly Dead
CLINICAL R A D I O L O G Y Table 3b - MRC 5-point scale of performance status
1. 2. 3. 4. 5.
Normal activity Fully active but not working Out and about but restricted activity Confined at home or hospital Bedridden
continue useful work and to care for his own needs, or his dependence on nursing or hospital care. Its use is exemplified by Hyde's study (1973) of inoperable lung cancer, with a consistent linear correlation between log-survival and initial Kamofsky rating which is not entirely surprising in this extensive disease group. In this country the categories are usually reduced to five, as in the MRC performance scale (Table 3b). The problems of function and quality evaluation are never more difficult than following the prolongation of life in patients with malignant brain tumours. Rosenblum and Walker (Rosenblum et al., 1973; Sklansky et al., 1974) have elaborated the Kamofsky system to take into account both neurological signs and symptoms (Table 4) and have described a Central Nervous System Disease Staging System in which a patient is placed in one or four disease stages based on a combined evaluation of symptoms, signs and function. Their criterion of therapeutic effect was, as already mentioned, improvement by one or more disease stages maintained at least until the end of a course of chemotherapy. 'Quality' of survival for such patients still eludes definition, and the philosophical overtones are revealed in Penman and Smith's (1954) term of 'useful survival', meaning that in their view 'a sensible on-looker, knowing and bearing in mind the sort of person the patient was before his illness, would on the whole be glad he was still alive'. The affective Table 4 - CNS disease staging system of Rosenblum and Walker (from Sklansky et al., 1974)
A. Signs Mental status Speech Visual fields Intracranial hypertension Cranial nerves
Motor Sensory Cerebellar Other
Graded from 1 (normal) to 4 (severe deficit)
B. Symptoms and function Disease stage = highest number in either category A or B.
changes which are part of the neurological deficit usually prevent such patients from themselves taking part in a subjective assessment; but in advanced breast cancer this is not so, and Priestman and Baum (1976) have pioneered the technique of 'linear analogue self. assessment' (LASA) to measure the subjective effeet~ of treatment. LASA is a standard form of psyche. logical testing which they adapted using indices SUch as well-being, activity, pain (and including the question 'is treatment helping?') so as to produce a numerical score. Applied to both response and toxi. city it gave in their hands reliable and reproducible results, and while it gives of necessity a superficial view, it is a first step towards recognising and quanti. tating the supreme importance of quality of life in assessing the overall response to chemotherapy. PERSONAL END-POINTS
What of the quality of life of patients who are regarded as 'cured', and have been told so by their physician? Each will have his or her own 'end-point' of hopes and aspirations, even of assumptions, which it is the place of the physician to support. But with any evaluation of 'quality' we pass straight into the realm of philosophy. It has been said that for a life of fulfilment a person needs three things: something to do, someone to love and something to look forward to: and perhaps I may be forgiven for looking at these in the present context. As by modern standards mere survival is quite inadequate as an end-point, any criterion for quality must also be set in contemporary terms. Something to Do
The employability of those cured of cancer has been examined several times, most recently by a large New York organisation, the Metropolitan Life Insurance Company (Wheatley et aL, 1974); it has found that about 50% of applicants with a cancer history (and who had had a wide variety of tumours) could be medically accepted for employment, and that their subsequent employment record was saris. factory. Community and industrial health physicians have their part to play in ensuring that patients ha this category are not placed at a disadvantage when being considered for employment, for even the NHS itself has been known to show reluctance to take on what the uninformed regard as 'yet another lame duck'. Someone to Love
As we have been exploring quantitation I hasten to reassure you, Mr President, that it is no part of this lecture to try to quantify amatory pursuits. But ha
E N D - P O I N T S IN C A N C E R T H E R A P Y
keeping with the theme, the objectives must be related to contemporary ideas and expectations, and attention has to be directed both to emotional needs ~ad to the endocrinological and procreative aspects. If we take these in the reverse order, a great deal is becoming known about the fertility of patients cured of cancer, and the impetus has come mainly from the studY of Hodgkin's disease. Now that about 80% of pathologically staged patients in Stages 1-3 can expect to remain symptom-free after radiotherapy, the reproductive implications for those receiving abdominal irradiation have become important. In an attempt to preserve ovarian function in young women so treated, oophorpexy has been performed routinely at the time of surgical staging, the ovaries being transposed to the midline to be later shielded by 10cm lead blocks during inverted-Y megavoltage irradiation. Reporting on an original series from Stanford, Le Floch et al. (1976) found that twothirds of the women retained ovarian function, and six had given birth to eight babies who showed no abnormality. Thomas et al. (1976) reporting on a series from the Royal Marsden Hospital investigated both reproductive and endocrine function, and found that the procedure of bilateral midline oophorpexy did not impair ovarian function or gamete transport: they considered that it should be performed at staginglaparotomy in young women even when it was uncertain whether pelvic node irradiation would be necessary. Of greater import is what de Vita termed the 'surprisingly high risk of functional castration in women receiving chemotherapy for lymphoma' (Sherins et al., 1975). Their study revealed that in women who had six courses of MOPP and who were reviewed one to nine years later, 57% had persistent amenorrhoea. Certain questions immediately arise: to what extent may the neoplasia itself indirectly determine infertility? Given that a degree of infertility is likely to occur in either male or female after chemotherapy, when does this happen and at what dose, and after how many courses? What gonadal structures are damaged (is it only a germinal cell depletion)? Does recovery occur, and can it be predicted for the individual patient? Such questions are likely to be answered very soon by the endocrinological studies now in progress (such as those of Chapman and her colleagues), and these may have implications for the management of the individual patient which extend beyond the provision of gamete storage facilities by 0ncological units. While such considerations must not tnfluence the use of chemotherapy necessary for cure, they may modify its application in certain adjuvant situations.
At the endocrinological and emotional levels the implications for management are considerable in the context of the modern orthodoxy of the Western world. While young patients accept generalisations on prognosis qua neoplasia, they now expect detailed discussion of the implications of chemotherapy. In the light of series experience, and of individual hormonal estimations, the patients with their spouses will need counselling on their expectations, pari passu with any hormonal replacement indicated by these studies. Something to Look Forward to
Apart from the hopes of the individual centred on family and work, the aspirations of modern society are firmly tied to financial prosperity, and here life insurance has come to play an increasingly important role. It is also here that the 'cured' patient, in youth or early middle age and about to purchase a house or to be promoted in his firm, may experience his first set-back: for he has been encouraged to regard himself as cured and he is perplexed when an insurance company seems to think differently. While in the past underwriting policies were based on historical concepts of cancer mortality which were of little relevance to the individual patient, there is at last a new look on this scene, exemplified by Brackenridge's (1977) perceptive and enlightened work which will repay study by all oncologists advising such patients. If physicians will provide the maximum amount of essential information, a realistic appraisal can be made: and in this respect there may be an additional bonus for patients who have had detailed staging procedures such as lymphography and full histopathological assessment of excised tumour and lymph nodes. Those interested will Fred that important touch-stones - seminoma, Hodgkin's disease and well-differentiated thyroid carcinoma - can be dealt with in an enlightened manner, but that breast carcinoma is still loaded with extra premium even when the survival is parallel to that of the normal population: which is just about where we came in. CONCLUSION As we have progressed from considering the control of neoplasia in groups to the care of the individual, we have seen the importance of attempting strict definition of our terms. While the experimentalist may get along with defining response as 'the particular measure of efficacy that is germane to the experiment at hand' (Miller, 1974) the clinician must bear in mind the connotation which his use of a term has for his colleagues in other branches of oncology.
Concepts of 'cure' were originally devised so as to reassure patients, but one may suspect that some recent extrapolations have been derived equally to reassure the oncologists themselves. If they are not on sure ground when communicating within clinical science the issues may rebound, for there are those who, in relation to the funding of cancer research, do n o t hesitate to point out that it was 'the cancer researchers who actually originated the numbers game'. We have seen that there is no single criterion of response which can be used in all neoplastic diseases, nor at all stages of the same disease, and the important feature is the change in pattern o f relevant parameters. Pattern recognition, both conscious and subliminal, has probably always been the basic method of clinical medicine, and in evaluating the results of cancer treatment we are, with all modern aids, just extending this basic medical expertise. The natural history of many cancers has been so modified by treatment as to require a change in endpoints, and with fuller control of neoplasia we are now concerned with matters which even a decade ago would have seemed academic. Undoubtedly the quality o f survival is of a different order of importance from the effects which are measurable, and it is equally the most difficult to evaluate. The energies of clinical oncologists will in future be increasingly devoted to considering this paramount feature of any therapeutic equation, and if it is not assessable in comparative statistical terms, this is a characteristic shared with all the highest activities of man. Acknowledgements. I wish to express my thanks to the following for kindly permitting reproduction of illustrations: Dr Ray Bush for Figs 1 and 2; Drs Brinkley and Haybittle for Fig. 3; Dr E. D. Thomas for Fig. 4; and Professor G. M. Besser and Dr J. Wass for the use of unpublished data in Figs 6 and 11 respectively.
Alberts, D. S., Durie, B. G. M. & Salmon, S. E. (1976). Doxorubicin/BCNU chemotherapy for multiple myeloma in relapse. Lancet, 1,926-928. Alexanian, R. (1977) Plasma cell neoplasms. Medicine, 2nd Series, ed. Malpas, J. S. & Weatherall, D. J., 30, 17001712. Bagshawe, K. D. (1969). Choriocarcinoma. Arnold, London. Bagshawe, K. D. (1976). In Cancer Related Antigens, ed. Franchirmont, P., p. 251. Elsevier Amsterdam. Barendsen, G. W. & Broerse, J. J. (1969). Effects of 15 MeV neutrons and 300 kV X-rays on a rat rhabdomyosarcoma. European Journal of Cancer, 5,373-391. Brackenridge, R. D. (1977). Medical Selection of Life Risks, p. 486. Undershaft Press, London. Brinkley, D. & Haybittle, J. L. (1975). The curability of breast cancer. Lancet, 2, 95-97.
Burgess, C. A. & Edwards, C. R. W. (1978). Hirsutography British Journal of Photography, 125,770-772. Bush, R. S., Gospodarowicz, M., Sturgeon, J. & Alison,1~ (1977). Radiation therapy of localised non-Hodgkt~, lymphoma. Cancer Treatment Reports, 61, 1129-1136 Bush, R. S. (1979). Malignancies of the ovary, uterus and cervix. In The Management of Malignant Disease Sene~ ed. Peckham, M. J. & Carter, R. L. Edward Arnold, London. Carter, S. K. (1977). Clinical trials in cancer chemotherapy Cancer, Suppl., 40, 544-557. Dawson, J. M., Hall, T. C., Schneiderman, M. A., Schnider, I., Owens, A. H., Anderws, J. R., Baxter, D. H., Brenner S., Hunter, C., Levene, M. B., Sheenan, F. R. &White,Gi (1966). Objective evaluation of change in turnout sizeaq lung cancer patients with non-measurable disease. Cancer, 19,415-420. Easson, E. C. (1966). Possibilities for the cure of Hodgkin's disease. Cancer, 19, 345-350. Easson, E. C. & Russell, M. H. (1963). The cure of Hodgkin's disease. British Medical Journal, 1, 1704 - 1707. Ell, P. J., Williams,E. S. & Todd-Pokropek, E. E. (1978). ~e clinical use of diagnostic imaging. British Journal o[ Hospital Medicine, 20, 119-127. Engstrom, J. E. & Austin, D. F. (1977). Interpreting cancer survival rares. Science, 195, 847-851. EORTC Chnical Screening Group (1973). Epipodophyll0. toxin VP 16213 in treatment of acute leukaemias, haerna. tosarcomas, and solid tumours. British Medical Journal, 30, 199-202. Frei, E. & Gehan, E. A. (1971). Definition of cure for Hodgkin's disease. Cancer Research, 31, 1828-1833. Gehan, E. A. & Freireich, E. J. (1974) Non-randomised controls in cancer clinical trials. New England Journal of Medicine, 290, 198-203. Greenwood, M. (1926) A report on the natural duration of cancer. Ministry of Health Report No. 33 HM Stationery Office, London. Hayward, J. L., Rubens, R. D., Carbone, P. P., Heuson. J. C, Kumaoka, S. & Segaloff, A. (1977). Assessment of response to therapy in advanced breast cancer. British Journal of Cancer, 35,292-298. Hill, R. P. & Bush, R. S. (1969). A lung-colony assay to determine the radiosensitivity of the cells of a solid tumour. International Journal of Radiation Biology, 15, 435-444. Hyde, L. Wolf, McCracken, S. & Yenser, R. (1973). Natural course of inoperable lung cancer. Chest, 64, 309-312 Karnofsky, D. A. & Burchenal, J. H. (1948). In Evaluation of Chemotherapeutic Agents, ed. MacLeod, C. M., London. Le Floch, O., Donaldson, S. S. & Kaplan, H. S. (1976) Pregnancy following oophoropexy and total nodal lrradla. tion in women with Hodgkin's disease. Cancer, 38, 22632268. Leiberman, D. P. (1978). CRC Breast Study Annual Meeting Report (personal communication). McNally, N. J. & Sheldon, P. W. (1977). The effect of radiation on tumour growth, cell survival and cure of the animal using a single tumour system. British Journal of Radiology. 50, 321-328. McNally, N. J. (1975). The effect of an hypoxic cell sens~tlser on tumour growth delay and cell survival.British Journal of Cancer, 32,610-618. Miller, I. (1974). Statistical design for experiments in conabl" nation therapy. Cancer Chemotherapy Reports, 4, 151156. Moertel, G. C. & Hanley, J. A. (1976). The effect of measur"
END-POINTS IN C A N C E R T H E R A P Y l~g error on the results of therapeutic trials in advanced cancer- Cancer, 38, 388-394. g,rraY J' G. (1974). Cancer Research Campaign Breast ~studY' British Journal of Surgery, 61,772-774. Neville, A. M. & Cooper, E. H. (1978). Biochemical monitoring of cancer. Annals of Clinical Biochemistry, 13, 283-305. po~man, J. & Smith, M. C. (1954). Intracranial gliomata. ~l~edical Research Council Special Report Series, No. 284, p. 66. Feto, R., Pike, M. C., Armitage, P., Breslow, N. E., Cox, D. R., Howard, S. V., Mantel, N., McPherson, K., Peo, J. & smith, P. G. (1976). Design and analysis of randomised clinical trials requiring prolonged observation of each patient. I. Analysis and design. British Journal of Cancer, 34,585-612. peto, R., Pike, M. C., Armitage, P., Breslow, N. E., Cox, D. R., Howard, S. V., Mantel, N., McPherson, K., Peo, J. & smith, P. G. (1977). Design and analysis of randomised clinical trials requiring prolonged observation of each patient. II. Analysis and examples. British Journal of Cancer, 35, 1-39. phillips, T. L. & Fu, K. K. (1976). Quantitation of combined radiation therapy and chemotherapy effects on critical normal tissues. Cancer, 37, Suppl. 2, 1186-1200. powles, T. J., Leese, C. L. & Bondy, P. K. (1975). Hydroxyproline excretion in patients with breast cancer and response to treatment. British Medical Journal, 2, 164166. Pnestman, T. J. & Baum, M. (1976). Evaluation of quality of life in patients receiving treatment for advanced breast cancer. Lancet, 1,899--902. Puck, T. T. & Marcus, P. I. (1956). Action of X-rays on mammalian cells. Journal of Experimental Medicine, 103, 653 -667. Rosenblum, M. L., Reynolds, A. F., Smith, K. A., Bumack, B. H. & Walker, M. D. (1973). CCNU in the treatment of malignant brain tumours. Journal of Neurosurgery, 39, 306 -314. Russell, M. H. (1958). Contribution to Symposium on Presentation of Results of Cancer Treatment, London. 7th International Cancer Congress. Salmon, S. E. (1973). Immunoglobulin synthesis and tumour kinetics of multiple myeloma. Seminars in Haematology, 10,135 147. Salmon, S. E., Hamburger, A. W., Soehnlen, B., Durie, B. G. M., Alberts, D. S. & Moon, T. E. (1978). Quantitation of differential sensitivity of human turnout cells to anticancer drugs. New England Journal of Medicine, 298, 1321-1327. Schaner, E. G., Head, G. L., Doppman, J. L. & Young, R. C. (1977). Computed tomography in the diagnosis, staging
and management of abdominal lymphoma. Journal of Computer Assisted Tomography, 1, 176-180. Shearer, R. J., Constable, A. R., Girling, M., Hendry, W. P. & Fergusson, J. D. (1974) Radioisotopic bone scintography with the gamma camera in the investigation of prostatic cancer. British Medical Journal, 2, 362-366. Sherins, R., Winokur, S. & De Vita, V. T. (1975). Surprisingly high risk of functional castration in women receiving chemotherapy for lymphoma. Clinical Research, 23, 343A. Skipper, H. E., Schabel, F. M., Jr & Wilcox, W. S. (1964). Experimental evaluation of potential anti-cancer agents. XIII. On the criteria and kinetics associated with 'curability' of experimental leukemias. Cancer Chemotherapy Reports, 35, 1-111. Sklansky, B. D., Mann-Kaplan, R. S., Reynolds, A. F., Rosenblum, M. L. & Walker, M. D. (1974). 4'-Dimethyl-epipodophylotoxin-#-D-thenylidene glucoside (PT6) in the treatment of malignant intracranial neoplasms. Cancer, 33,460-467. Steel, G. G. (1977). Growth Kinetics of Tumours, p. 307. Oxford. Sutcliffe, S. B., Wrigley, P. F. M., Peto, J., Lister, T. A., Stansfeld, A. G., Whitehouse, J. M. A., Crowther, D. & Malpas, J. S. (1978a). MVPP chemotherapy regimen for advanced Hodgkin's disease. British Medical Journal, 1, 679-683. Sutcliffe, S. B., Stansfeld, A. G., Wrigley, P. F. M., Katz, D., Shand, W. S. & Malpas, J. S. (1978b). Post-treatment laparotomy in the management of Hodgkin's disease. Lancet, 1, 57-60. Thomas, E. D., Flournoy, N., Buckner, C. D., Clift, R. A., Fefer, A., Neiman, P. E. & Storb, R. (1977). Cure of leukaemia by marrow transplantation. Leukaemia Research, 1, 67-70. Thomas, P. R. M., Winstanly, D., Peckham, M. J., Austin, D. E., Murray, M. A. F. & Jacobs, H. S. (1976). Reproductive and endocrine function in patients with Hodgkin's disease: effects of oophoropexy and irradiation. British Journal o f Cancer, 33, 226-231. Thomlinson, R. H. (1977). Implications of the changing size of tumours. British Institute of Radiology Symposium, December 1977 ; and personal communication. Wara, W. M., Phillips, T. L., Sheline, G. E. & Schwade, J. G. (1975). Radiation tolerance of the spinal cord. Cancer, 35, 1558-1562. Wheatley, G. M., Cunnick, W. R., Wright, B. P. & van Keuren, D. (1974). The employment of persons with a history of treatment for cancer. Cancer, 3 3 , 4 4 1 - 4 4 6 . Wiltshaw, E. & Kroner, T. (1976). Phase II study of cisdichlorodiammineplatinum II (NSC-119875) in advanced adenocarcinoma of the ovary. Cancer Treatment Reports, 60, 55-60.