| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Commentary |
1 Radiology Department, Great Ormond Street Hospital For Children, London WC1N 3JH, UK and 2 Department of Radiology, The University of Iowa College of Medicine, Iowa City, IA 52242, USA
Correspondence: Kieran McHugh
The new Response Evaluation Criteria In Solid Tumours (RECIST) guidelines have been formulated to document tumour size change and monitor treatment response in oncological imaging [1]. From 1979 the World Health Organization (WHO) handbook popularized four specific criteria for the codification of response evaluation in solid tumours [2]. The four categories, namely, complete response (CR), partial response (PR), stable disease (SD) (also termed "no change"), and progressive disease (PD) came to be used widely both in adult and paediatric oncology practice. Four major problems with these definitions gradually evolved over time: 1) the methods of integrating the change in tumour size into response assessments varied among research groups; 2) the minimum lesion size and number of lesions to be documented also varied; 3) what constituted progressive disease was based on the change in size of a single lesion by some researchers and a change in the overall tumour load (including measurements of all lesions) by others; and 4) new technologies, particularly CT and MRI further confused matters as regards the relevance of volumetric and three-dimensional (3D) measurements in response assessments [3]. Consequently, a situation arose whereby the response criteria were no longer comparable among research organizations – the very circumstance that the original WHO publication had set out to avoid [1]. For these reasons, the WHO, the European Organisation for Research and Treatment of Cancer, the National Cancer Institute of the United States and the National Cancer Institute of Canada Clinical Trials Group set up a task force in 1994 with the main objectives of reviewing the existing sets of criteria used to evaluate response to treatment in solid tumours. After 3 years of regular meetings, a revised version of the WHO criteria was produced and circulated [1]
.
|
The RECIST document was created as a result of the discussions mentioned above. In essence the document has two parts, the RECIST criteria and the RECIST guidance (with specific imaging recommendations). Most of our comments here relate to the RECIST guidance. The rationale behind RECIST was to unify response assessment, to try to better define evaluable lesions in patients with solid tumours and to take into account new imaging techniques [5]. The RECIST guidelines continue to rely on size change of lesions to make response assessment. The four response categories (CR, PR, SD and PD) have been retained to enable comparison of past and future treatments. Under RECIST, the methods by which lesions are assessed has changed, however. Unidimensional measurement of maximum diameter rather than a bidimensional measurement of lesion area is now employed. Unidimensional measurements were chosen because of the relationship between change in diameter, product and volume of spherical lesions – and for reasons of simplicity [4]. Precision in measurement was deemed less important than standardized methodology. Despite a strong intuitive feeling that two measurements should be better than one, the new criteria have proven to be reliable and equivalent to bidimensional measurements in large numbers of adults [4]. Diameter is supposedly more nearly proportional to the logarithm of cell number than product and so changes expressed in units of diameter are roughly independent of the initial tumour sizes in different patients [4]. Diameter, however, becomes grossly inaccurate as an estimate of tumour size when the length of a mass lesion is more than twice its width, as would often be the case with a paraspinal neuroblastoma [6, 7]. There are practical reasons nevertheless why a single diameter should be chosen over the product of two diameters perpendicular to each other [4]. There is a saving in calculation and the simplicity of measurement encourages more lesions to be measured in an individual patient. It is generally accepted that the greater the number of lesions measured, the lesser chance there is of falsely deciding that a partial response has occurred [3]. Although the recommended unidimensional measurement by RECIST could be described as a simplified way of measuring one or more lumps, critical clinical decisions are made daily based on the results of these measurements.
While the four categories of response assessment have been maintained there has been slight alteration in their determination. CR and SD remain the same under RECIST as with the older WHO system. It was agreed that no major discrepancy in the meaning and the concept of partial response (PR) should exist between the old and the new guidelines, but measurement criteria would be different [1, 5]. To achieve PR now at least a 30% decrease in the sum of the longest diameters must be evident. This is roughly equivalent to the previous WHO 50% reduction in the product of the diameters [1, 5]. These criteria are equivalent if one assumes tumours have a spherical shape and that the longest diameter (LD) and the diameter perpendicular to the LD both decrease by at least 30% (although the latter was not measured in the RECIST studies) [8]. The only truly significant change in definition has occurred with the PD category, whereby now greater volume increases are necessary to define disease progression. The RECIST criteria require a 20% increase in diameter for a single lesion or in the sum of the LDs of multiple masses to define PD. The need for greater tumour burden has come about as concerns had been raised regarding the ease with which a patient could have been mistakenly considered to have disease progression by the WHO criteria, and in an attempt to minimize the contribution of enlargement of small lesions [1, 5].
The RECIST guidance defines in some detail whether tumour lesions may be classified as measurable or non-measurable, target or non-target lesions. These concepts and terms are new to imaging in paediatric oncology. They probably have greatest applicability in the setting of adult phase II clinical trials of metastatic malignancies. A measurable lesion is defined as one which can be accurately measured in at least one dimension, and which should measure 10 mm or more on a spiral CT examination. Non-measurable lesions are lesions smaller than 10 mm, but also include (dubbed "truly" non-measurable) bone metastases, leptomeningeal secondaries, malignant ascites, pleural or pericardial effusions, lymphangitis and heavily calcified or cystic/necrotic lesions [5]. It is noteworthy here that many solid tumours in children are cystic or necrotic at first presentation and these apparently necrotic masses do frequently shrink with chemotherapy nonetheless. (Such necrotic masses are more likely pseudonecrotic which is why they shrink with treatment). Therefore cystic or necrotic masses should be designated as measurable in paediatric studies. Conversely, cystic portions of some masses may on occasions actually increase in size with chemotherapy despite no viable tumour being found after resection. It could thus be argued that both the solid and cystic portions of tumours, either singly or in combination, should routinely be measured although that is likely to prove difficult to implement in practice. After establishing that measurable disease is present, it is necessary to document target and non-target lesions. Five measurable lesions per organ and 10 measurable lesions in total, representative of all involved organs, should be documented as target lesions. The sum of the diameter of all target lesions constitutes the "baseline sum longest diameter". All other lesions (or sites of disease) should be identified as non-target lesions and should also be recorded at baseline. Measurements of these non-target lesions are not required but their presence or absence should be noted throughout follow-up. Non-target lesions could thus be in the measurable range (>10 mm) or non-measurable.
CR and SD are possible with non-target and target lesions. With non-target lesions PR is not possible (as they are not measured) but PD may occur if one or more new lesions are seen or there is unequivocal progression of existing non-target lesions. A potential pitfall with RECIST is that these criteria do not take into consideration differences in response, and perhaps prognosis, according to which metastatic organs are affected [9]. Neuroblastoma in infancy, for example, may be metastatic to liver, skin and bone marrow (stage 4S disease which has a very good prognosis) or to bone cortex and elsewhere (stage 4 disease with a very different outcome). This pitfall is not, of course, unique to RECIST. It is primarily a limitation of solely using size criteria in response evaluation.
The specific recommendations, relating to CT in particular, contained within the RECIST guidelines pose a major problem for paediatric patients. There is excess reliance on CT which is unfortunate as the high inherent radiation burden from CT precludes its repeated use in children. It is alleged that "the lifetime cancer mortality risk attributable to the radiation exposure from a single abdominal CT examination in a 1-year-old child is of the order of one in 1000" [10]. Such an hypothesis may be impossible to prove and the radiation dose, of course, is dependent on the acquisition parameters used. Nevertheless, CT is now recognised as a dangerous modality if used excessively in children [10, 11]. For these reasons, CT should be used sparingly, even in young patients with cancer, as many have an excellent long-term outcome. The CT protocols within the RECIST guidance are particularly detailed and apparently prescriptive. The minimum lesion size to be measured is recommended by RECIST to be double the slice thickness. With a conventional CT scanner 20 mm minimum diameter is actually the stated lesion size to be measurable. This is based on the 10 mm thick sections commonly used in adult CT scanning when screening for metastases. This is unlikely to be a particular problem for paediatric radiologists as the majority of paediatric oncology centres will have access to spiral CT machines capable of providing subcentimetre slice thickness. 5 mm sections, routine for scanning the chest or abdomen in most young patients, allow 10 mm masses to be measured. Others have found that 10 mm lesions are indeed evaluable in the RECIST context [12]. The categorization of all lesions less than 10 mm (spiral CT) or 20 mm (conventional CT) as unmeasurable may be unrealistic and presents a problem in that the number of target lesions in an individual patient may be reduced [9]. Further modification allowing for smaller tumours to be measurable should now be possible with spiral or newer multislice CT scanners. For example, 2–4 mm sections may be performed to detect a pituitary lesion or a small, but obvious, posterior fossa recurrence can easily be measured where the tumour is at least twice the slice thickness but less than 10 mm in size.
RECIST stipulates that the LD of each target lesion should be selected on CT in the axial plane only [1]. When MRI is used, lesions must be measured in the same anatomic plane by use of the same imaging sequences on subsequent examinations [1]. There may, however, be some discrepancy in classification of tumour response when the axial diameter only rather than all three axes (x, y, z) are used in the selection of the LD [13]. Bidimensional measurements can provide a more accurate assessment of treatment response than diameter alone in tumours which develop into shapes such that their width is more than twice the length, as may occur on chemotherapy [14]. With shrinkage a mass may in addition alter its orientation such that the LD is no longer in the same orthogonal plane as on the previous study. For these reasons some paediatric radiologists may prefer to measure a mass in all three anatomic planes until RECIST has been validated to provide an accurate assessment of size change in paediatric tumours.
RECIST recommends that the same MR scanner wherever possible be used when following up patients. The reality, however, is that some children get initially worked up in a peripheral hospital and are then transferred to a paediatric oncology centre. Follow-up studies cannot be done therefore on the same or similar strength scanner. This also applies to CT scanning, such that repeating a study with exactly the same windows and level may not be possible, although similar windows and levels are strongly recommended. Children with localized, favourable prognosis abdominal tumour such as a Wilms' mass require a combination of ultrasound and CT or MRI at diagnosis and thereafter most follow-up could be with ultrasound, or ultrasound alternating with occasional CT. RECIST will somehow need to acknowledge this in the paediatric setting. Similarly, the anxiety surrounding a new paediatric tumour is such that urgent CT to document for example an orbital rhabdomyosarcoma, is likely to remain common, but scheduled routine follow-up should be permissible with MR, with measurements best done for comparison in the same axial plane as on the original CT. It may thus be acceptable in specific circumstances to interchange modalities, which is contrary to RECIST guidance [1]. If some interchange of imaging modalities is permitted, one must, however, bear in mind that accuracy of tumour size estimation may necessarily diminish. A mass measured on CT may not be the same size on either T1 weighted or T2 weighted MRI owing to differences in principles of detection (attenuation versus signal intensities, blood, calcium etcetera).
In an ideal world virtually all paediatric solid tumours would be scanned with MRI combined with ultrasound where appropriate and CT would no longer be necessary. The use of ultrasound is discouraged by RECIST as ultrasound is an operator dependent technique. In addition ultrasound examinations cannot be easily reproduced for independent review at a later date. It has already been pointed out in this journal that clinical palpation, endoscopic evaluation and even pathological evaluation are also highly operator dependent [5]. Ultrasound is permitted as an alternative to clinical measurements for thyroid nodules, subcutaneous lesions and superficial palpable lymph nodes. It could be argued that all mass lesions in children are superficial (bearing in mind over 50% of paediatric cancers occur in those less than 5 years of age) and thus amenable to ultrasound measurement. Suffice here to say that RECIST guidance will need to be interpreted flexibly in the paediatric setting to take into account everyday clinical practice in paediatric oncology. Ultrasound is the modality best suited to repeated scanning of abdominal, and many other, masses in small children – avoiding the hazards of repeated irradiation, sedation and general anaesthesia. Moreover, RECIST recommends that the same imaging modality be used to characterize each identified lesion at baseline and during follow-up, thereby adding further weight to the need to be able to rely on ultrasound evaluation in paediatric patients. If ultrasound is permitted to measure paediatric masses under a modified RECIST system, to aid reproducibility we recommend measurements in three orthogonal axes should probably be used at baseline and follow-up. Use of the standardized planes, namely axial, sagittal, and coronal, should decrease variation between studies. Unless this is emphasized, significant intraobserver and interobserver variability is likely to be encountered.
Although an improvement on the old WHO evaluation process, there are still occasions when RECIST has limited applicability. Situations exist where no tumour shrinkage may be evident on radiological follow-up but a clear histological response can be seen, e.g. rhabdomyomatous change in bilateral Wilms' tumour [15, 16]. Conversely, tumour progression can be inferred when a rise in a tumour marker occurs, e.g. rising alphafoetoprotein levels in a child with hepatoblastoma, despite no objective evidence on imaging of tumour size change. It is becoming increasingly clear from nuclear medicine functional imaging, notably positron emission tomography, that metabolic and physiological changes antecede tumour size change [17]. Newer cytostatic or non-cytoreductive agents, such as gleevec (imatinib mesylate), which are not anticipated to produce tumour shrinkage will require different end points to identify drugs of promise for evaluation in larger trials. Although nuclear medicine's contribution to staging of malignancies is essentially ignored by RECIST (the criteria are primarily concerned with tumour size change), metaiodobenzylguanidine (MIBG) scanning, for example, will remain crucial to accurate staging and follow-up of metastatic neuroblastoma for the foreseeable future.
Current RECIST guidance states that the frequency of tumour re-evaluation while on treatment should be protocol specific and adapted to the type and schedule of treatment [1]. Although it is recommended that CR and PR should always be confirmed at a further study 4 weeks later, the need to confirm response would only need to be adhered to closely in the context of trials of pharmaceutical agents. It is noteworthy that much of the early RECIST discussions took place in the early 1990s and so the reliance on CT is understandable in that context. In children with metastatic tumours, the suggested frequency of repeated CT is not really an issue as the long-term survival of most childhood metastatic malignancies remains poor. In good prognosis tumours such as localized abdominal Wilms' tumour, however, repeated CT at 4-weekly intervals would not be justifiable. Not all protocols demand 4-weekly follow-up, but there can be a requirement to confirm CR or PR by repeating a study 4 weeks later. Here a greater reliance on ultrasound will be necessary.
A perfect system for measuring all the different paediatric malignancies in both the brain and elsewhere is difficult to envisage for a number of reasons. Outcomes and treatment in paediatric oncology are constantly changing. MRI (or ultrasound or CT) may not be the best modality to objectively document tumour size change 20 years from now. New functional imaging parameters that reflect tumour vascularity and cell metabolism may soon need to be taken into account as tumour response variables [16]. The paradigm of radiological assessment of drug efficacy remains particularly attractive in oncology, however, because tumour size can be monitored easily and serially [14]. Hence the RECIST criteria are likely to stay for the foreseeable future.
When the RECIST document was being formulated, there was a regrettable lack of consultation with the paediatric radiological and oncological community. The guidelines with regard to CT need to be modified for paediatric oncology to take into consideration practical issues such as repeated (unnecessary or too frequent) irradiation, and also a potentially excessive need for sedation and anaesthesia. The increased demand for imaging in general will also have cost and manpower implications. Although the same method of assessment and the same technique being used at baseline and follow-up is ideal, in practice CT at diagnosis followed by non-emergent MRI or ultrasound may have to be allowed. If some interchange of imaging modalities is permitted, it should be borne in mind that accuracy of tumour size estimation will necessarily diminish. Nevertheless a combination of either CT or MRI plus ultrasound may be used for, say, abdominal masses with repeat assessments and measurements, with ultrasound. Most significantly, although discouraged by the RECIST guidance, ultrasound is the main imaging modality in paediatrics and its general reliability will need to be officially accepted as a real difference between adult and paediatric oncology practice. Finally, the accuracy of tumour volume estimations is not always crucial in an individual patient and the significance of functional imaging, e.g. MIBG, should not be forgotten.
Received for publication November 22, 2002. Revision received January 28, 2003. Accepted for publication April 24, 2003.
References
This article has been cited by other articles:
 |
|
 |
|
  C. H. Takimoto Commentary: Tumor Growth, Patient Survival, and the Search for the Optimal Phase II Efficacy Endpoint Oncologist, October 1, 2008; 13(10): 1043 - 1045. [Full Text] [PDF]
|
|
|
|
|
|
S. M. Schuetze, L. H. Baker, R. S. Benjamin, and R. Canetta Selection of Response Criteria for Clinical Trials of Sarcoma Treatment Oncologist, April 1, 2008; 13(suppl_2): 32 - 40. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. C. Jaffe Measures of Response: RECIST, WHO, and New Alternatives J. Clin. Oncol., July 10, 2006; 24(20): 3245 - 3251. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. Galanis, J. C. Buckner, M. J. Maurer, R. Sykora, R. Castillo, K. V. Ballman, B. J. Erickson, and for the North central cancer treatment Group Validation of neuroradiologic response assessment in gliomas: Measurement by RECIST, two-dimensional, computer-assisted tumor area, and computer-assisted tumor volume methods Neuro-oncol, April 1, 2006; 8(2): 156 - 165. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. T. Zacharia, S. Saini, E. F. Halpern, and J. E. Sumner CT of Colon Cancer Metastases to the Liver Using Modified RECIST Criteria: Determining the Ideal Number of Target Lesions to Measure. Am. J. Roentgenol., April 1, 2006; 186(4): 1067 - 1070. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L Moon and K McHugh Advances in paediatric tumour imaging Arch. Dis. Child., June 1, 2005; 90(6): 608 - 611. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. J. Kelloff, J. M. Hoffman, B. Johnson, H. I. Scher, B. A. Siegel, E. Y. Cheng, B. D. Cheson, J. O'Shaughnessy, K. Z. Guyton, D. A. Mankoff, et al. Progress and Promise of FDG-PET Imaging for Cancer Patient Management and Oncologic Drug Development Clin. Cancer Res., April 15, 2005; 11(8): 2785 - 2808. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| BJR | DMFR | IMAGING | ALL BIR JOURNALS |
