| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Full paper |
1Edinburgh University Department of Medical Physics, Western General Hospital, Edinburgh EH4 2XU and 2Quality Assurance Reference Centre, Newcastle General Hospital, Newcastle-on-Tyne NE4 6BE, UK
 |
Abstract |
|---|
 Top Abstract IntroductionNumbers of cancers detected...Cancers detected and induced...Subgroups of women receiving...ResultsDiscussionConclusionsReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionNumbers of cancers detected...Cancers detected and induced...Subgroups of women receiving...ResultsDiscussionConclusionsReferences |
|---|
Once the UK National Health Service Breast Screening Programme (NHSBSP) was well established, many of these numbers could be estimated with more confidence, although both benefit and risk from X-rays still remain relatively uncertain. Concern then began to arise over doses received by a very few women who required several additional films, and attempts were made to show that the probability of detecting an unsuspected existing cancer exceeded the probability of inducing a future breast cancer for all women in the programme. It was quickly realized that this could not be guaranteed without exception, but that the hypothetical proportion for whom that requirement could not be fulfilled was vanishingly small, at least within the age range of the NHSBSP of 50–64 years. Mammography remains the screening technique of choice.
In the last 5 years or so, cancer screening detection rates have continued to rise whilst X-ray doses and associated risk factors have shown little or no change. Various research studies have investigated extending the screening programme. Recently, the planned extension of the screening age range up to 70 years has been announced (for implementation by 2004), and two-view screening for all rounds is to be introduced by 2003. There is also increasing interest in screening of younger women (i.e. below 50 years of age), both with and without a family history of breast cancer. This paper attempts to re-assess the numbers of cancers detected and induced in the light of recent data. The relationship between these numbers and benefit/risk ratios will be explored in a separate paper. The emphasis will be on risks to the screened population as a whole and to defined higher dose subgroups within that population, rather than on individual women at the hypothetical highest doses.
|
Numbers of cancers detected and induced in the NHSBSP |
|---|
|
TopAbstractIntroductionNumbers of cancers detected...Cancers detected and induced...Subgroups of women receiving...ResultsDiscussionConclusionsReferences |
|---|
Cancer screening detection rates
In the NHSBSP, all women are invited at 3-yearly intervals. They receive two-view mammography in the first (prevalent) round and single view in subsequent rounds. It is intended to move to two-view screening on all rounds in the near future. Screening detection rates in the NHSBSP are reported at annual intervals. Those for the last 5 years for which figures are available are given in Table 1a
and refer to England only rather than the UK [1]. Data for other age ranges are given in Tables 1b,c. They refer to all screening rounds and do not distinguish between prevalent and incident rounds. They refer to all cancers, including ductal carcimona in situ (DCIS). There is a rising trend with time for the 60–64 year age band and that variation may be random (±3%), but there is a more pronounced rising trend for the two younger 5-year age bands. The increase has been more than 10% at age 55–59 years, and more than 20% at age 50–54 years. The increase in this last age band is probably partly explained by the gradual introduction of two-view screening on the first round, which occurred during 1994 and 1995. The mean values for 5 years given in Table 1a have been used in Table 4 and for subsequent calculations. A variation of up to approximately ±15% is found between different regions of the UK [2], and some of this is presumed to arise from geographical variations in the underlying breast cancer incidence. However, some of this apparent variation will be statistical.
|
|
|
). The mean dose per woman is slightly higher because a small minority of women have more than one film per view, especially where the larger film size of 24 cm x 30 cm is not available. The corresponding mean doses per woman for single view and two-view screening are 2.14±0.04 mGy and 3.65±0.07 mGy, respectively. These uncertainties are approximately 2%, representing 95% confidence limits, and are thus very small, especially in comparison with other uncertainties to be discussed later. The dose levels reported within the NHSBSP for 1997–8 agree with those found in similar work for 1994–5 within these small uncertainties [4].
|
Young and Burch [3] also consider the question of higher doses to the subgroup of women whose breasts are much thicker than average when compressed. Thicker breasts require higher radiation doses than smaller or average size breasts because a greater radiation exposure is required on the X-ray beam entrance side of the breast to ensure an adequate X-ray intensity on the exit side and onto the film, thereby maintaining an adequate optical density on the processed film. Failure to ensure this results in an inadequate image and potentially reduced cancer detection (i.e. more false negatives). Moreover, some breasts are so large as to require more than one film per view because they cannot be accommodated on a single 18 cm x 24 cm film. The availability of 24 cm x 30 cm films and cassettes is very helpful in reducing the number of exposures.
The highest doses are likely to be received by women with the thickest breasts (around 10 cm) who also require more than one film per view, possibly up to four such films. However, many surveys of breast thickness have shown that the number of women in this subgroup is extremely small and correspondingly difficult to predict. This point is further discussed later in this paper.
Radiation risk factors for breast cancer
The risk of radiation induction of breast cancer decreases with increasing age of the woman at the time of exposure [5]. Numerical values for the magnitude of this risk factor are based on work by the National Radiological Protection Board (NRPB), with further subdivision from 10-year to 5-year age bands following discussion with individuals from the NRPB [6]. Although this data set is 5 years old, these estimates of radiation risk factors are considered to remain the best available for the UK population. They refer to all breast cancers in a female population, not fatal cancers in a mixed population as in some earlier studies, and are given in Table 3. This table also gives corresponding cancer induction rates for the various dose levels for two-view screening, i.e. population averages, and 2% and 0.1% subgroups as described in a subsequent section of this paper.
|
|
Cancers detected and induced in screening of younger women |
|---|
|
TopAbstractIntroductionNumbers of cancers detected...Cancers detected and induced...Subgroups of women receiving...ResultsDiscussionConclusionsReferences |
|---|
Cancer screening detection rates in younger women without family history
These rates cannot be derived directly from existing programmes, which are currently based on small numbers and in most cases remain unpublished. A different approach is therefore required. Breast cancer incidence rates are known at all ages. Since cancer screening detection rates are well established at age 50 years and above, the rates at younger ages can be approximately estimated by simple proportion from the relative incidence rates, and then scaled proportionately for annual or 2-yearly screening. This procedure takes no account of the greater difficulty in cancer detection at younger ages arising from denser breasts and the associated problems of interpreting mammograms. However, at present most programmes for screening younger women are based on annual or 2-yearly screening. It is assumed in the calculations which follow that, after the first round, the screening detection rate with 2-yearly screening is two-thirds that for 3 yearly screening. A shorter interval than 3-years may be expected to reduce the number and proportion of interval cancers and thus lead to some proportionate increase in the detection rate thus calculated for 2-yearly screening. Although these two effects cannot be precisely quantified, they will tend to balance each other, and no attempt will be made here to correct for either. Although detection rates in the UK age trial (starting at age 40/41 years) remain unpublished, preliminary information on detection rates achieved suggests that the simple calculation described above will not be in any great error at least for ages 40–49 years. Even if detection rates were available from programmes for younger women, until these are based on larger populations, comparable in size with those screened at older ages, it may be safer to use calculations based on age-specific incidence, with a somewhat greater allowance for statistical uncertainty. The results of this approach are given in Table 1b
.
Cancer screening detection rates in younger women with family history
Akin to women without family history discussed in the previous section, detection rates cannot be derived directly from existing programmes. The reasons for this are the same as those already given for the former group. An alternative approach is provided by the work of Houlston et al [7], and this approach was used in an earlier paper [8]. It depends on the magnitude and time dependence of breast cancer incidence in first degree relatives (mother/daughter/sister) of those who have previously developed breast cancer at a given age, combined with cautious assumptions about the proportion of those cancers likely to be detected by screening. The results of this approach are given in Table 1c and have been taken unaltered from that earlier paper because no more reliable published data are known to us. However, results drawn and combined from a study involving 22 units in the UK show a detection rate (not stratified by age) of over 4.5 per 1000 for both prevalent and incident screening, based partly on annual screening and partly on 2-year screening (R D Macmillan et al, personal communication). Although this is based on a total of only 83 detected cancers, it quite strongly suggests that the values given in Table 1c are not overestimated.
MGDs for younger women
These might be expected to be somewhat greater than in women over the age of 50 years because younger breasts are often somewhat denser. No direct study of this point appears to have been published, but small-scale unpublished studies on women of different ages who have been screened using the same equipment suggest that down to age 40 years any increase is unlikely to exceed 30% and may be well below that figure [8]. This means that even if no allowance is made for this possibility, the consequent error is likely to be well within the other known uncertainties in the calculations. It is also interesting that Young and Burch [3] report no trend in MGD with age at 50–65 years.
|
Subgroups of women receiving higher doses |
|---|
|
TopAbstractIntroductionNumbers of cancers detected...Cancers detected and induced...Subgroups of women receiving...ResultsDiscussionConclusionsReferences |
|---|
Young and Burch [3] report that in their survey, 99.9% of oblique films had doses below 8.6 mGy and that in two-view screening the corresponding level was 15.7 mGy if one film per view was used. They suggest that these doses represent an upper limit to what can normally be expected, that only a small proportion of women receive doses at about this level and that this subgroup should be considered in any risk/benefit analysis. They add that a few women will receive doses higher than this, but this is very rare and impossible to predict in advance.
The essential principle of that approach will be adopted in this paper. Whether their suggested level of 0.1% as a proportion for whom higher doses should be disregarded will gain general acceptance remains to be seen. Some might argue for different numerical values, but their 0.1% level will also be adopted here.
The method of data acquisition used by Young and Burch did not enable them to identify those women whose films have to be repeated for technical reasons and on a separate occasion or repeat visit. One of the targets of the NHSBSP is to keep the proportion of such women below 3% of the total [9], a target that is being achieved in the majority of centres [10]. The consequent dose adjustment is, on a population basis, quite small, although for the individual women concerned the dose is doubled. Even for the 0.1% of the screened population receiving the highest doses (mainly those having the thickest breasts), the dose should not exceed 20 mGy, and this round number will make some allowances for statistical variation, since Young and Burch's 15.7 mGy is based on only a small number of women within their large sample.
One implication of the previous paragraphs is that a screening programme might aim to ensure that, within its own screened population, no more than 0.1% exceed 10 mGy (single view) or 20 mGy (two-view). Supposing that 20 mGy was the dose level at which risk exceeds benefit in a given age group, what action should then be taken if this 0.1% were exceeded by a small amount, for example 0.2%? Since the proportion is still very small compared with those likely to benefit, immediate suspension of screening would be inappropriate, but procedures or even equipment might be modified slightly provided that this did not adversely effect image quality and cancer detection rates. This then prompts the question: "at what proportion of screened women for whom risk did exceed benefit would screening be deemed to be unacceptable?" Selecting a numerical value for this is plainly a matter of judgement rather than of any precise calculation, but a range of 1–5% is suggested, and 2% will be used in tables as a basis for discussion. For two-view screening, this corresponds to a dose level of about 11 mGy.
|
Results |
|---|
|
TopAbstractIntroductionNumbers of cancers detected...Cancers detected and induced...Subgroups of women receiving...ResultsDiscussionConclusionsReferences |
|---|
. Ratios of numbers of cancers detected to numbers induced are shown in Tables 4–6 for women of screening age, younger women without family history of breast cancer and younger women with such a family history, respectively. In each of these tables, the three dose levels included are the average dose to the whole screened population, and the doses exceeded by approximately 2% and 0.1% of the screened population, as explained previously. These last two doses have been rounded to 11 mGy and 20 mGy, respectively, for reasons already indicated.
|
|
|
|
Discussion |
|---|
|
TopAbstractIntroductionNumbers of cancers detected...Cancers detected and induced...Subgroups of women receiving...ResultsDiscussionConclusionsReferences |
|---|
Dose levels in the NHSBSP have not changed by more than approximately 2% in the period between the last two major surveys. This constancy may not be maintained in the future, but allowing ±10% for the uncertainty in the current values is probably more than sufficient.
Uncertainties in the cancer induction risk factors are more difficult to quantify, and are rarely stated, but a factor of 2 in either direction is sometimes spoken of. This is then the dominant source in the total uncertainty in estimates of ratios of cancers detected to cancers induced, and an overall factor of 3 would seem more than sufficient. This implies that these ratios should always exceed 3 for all major screened groups and ages if we are to be confident that the true ratio will actually exceed 1.0.
However, the ratio of cancers detected to those induced is not necessarily the same as the true benefit/risk ratio. The relationship between these two ratios will be discussed in a separate paper, where it will be seen that the relationship depends both on the data used regarding treatment outcome and on whether women who have been in screening programmes continue to be screened (self-referred or otherwise) in later life. Given these uncertainties it would seem that at worst the benefit/risk ratio is at least half the ratio of detection to induction. It increases as more women are screened at ages over 64 years. It seems prudent to allow a further factor of 2 to take account of this point. Thus, if we are to be reasonably confident that benefit exceeds radiation risk in the NHSBSP, the ratio of detection to induction should exceed a factor of about 5 rather than 3. This may not apply to women below 40 years, with or without family history, because treatment benefit may be lower in this age group.
For younger women the uncertainties are somewhat greater, but not very much so. Detection rates are less well established and doses may be slightly higher, but the added uncertainty from these two sources combined is unlikely to reach a further factor of 2. Thus, cancers detected should exceed those predicted to be induced by a factor of about 10 in younger women.
A further point must be mentioned here that seems at present impossible to quantify. In recent years, radiation risks have tended to be calculated as a proportionate increase in the underlying cancer incidence in the exposed group, that is to say by the relative risk model rather than by the absolute risk model used previously. It is not certain that the relative risk model applies to breast cancer even if it is shown to be a better model for other cancers, but the possibility must be considered. If this relative risk model were also to apply to the increased underlying risk of those with a family history of breast cancer, then the results presented here for the family history group would be invalid. The true position for that group would then be represented, at least much more closely, by the results given for the no family history group. This possibility was raised in an earlier paper [8], but no further evidence on the point seems to have emerged since that date. If screening of family history groups, especially below the age of 35 years, is to proceed with any confidence, this point requires fairly urgent attention.
|
Conclusions |
|---|
|
TopAbstractIntroductionNumbers of cancers detected...Cancers detected and induced...Subgroups of women receiving...ResultsDiscussionConclusionsReferences |
|---|
imply that benefit is likely to exceed radiological risk by a substantial margin even at the dose level exceeded by only 0.1% of the screened population. The margin of benefit will be even greater at older ages, and would remain more than sufficient if the screening interval were reduced from 3 years to 2 years, or if the radiation dose were to (say) double to improve diagnostic accuracy.
For younger women without a family history of breast cancer, the margin of benefit over risk is sufficient at the average dose level for the screened group down to age 35 years, and at the dose level exceeded by 2% of the screened group down to age 40 years (Table 5). In the 35–39 year age band, on the basis of known statistical accuracy, risk may exceed benefit for more than 2% of the screened group, but not many more. It would be a matter for informed judgement whether that was acceptable were population screening to be offered in this age range. These remarks apply to annual screening. With a 2-year screening interval, comparable risks are reached for screening ages approximately 5 years younger.
For younger women with a family history of breast cancer in a first degree relative, and diagnosed at age 30–39 years, the margin of benefit over risk appears quite sufficient down to age 40 years at all three dose levels quoted in Table 6, provided always that the risk of radiation-induced breast cancer in this group remains similar to that of the general population. Down to age 35 years, the margin is sufficient at the dose level exceeded by 2% of the screened group, and down to age 30 years it is sufficient at the average dose level and at doses exceeded by a small proportion but more than 2% of the group. Screening at ages 25–29 years, however, should be treated with considerable caution.
Where the age of the relative at diagnosis is in the 40–49 year band, the margin of benefit is somewhat reduced at all ages of screening, so that the balance of benefit and risk is about the same for screening at 35–39 years as it is at 30–34 years in those whose relatives were diagnosed at 30–39 years. Screening in this subgroup at ages 30–34 years should also be approached with considerable caution. As before, these remarks apply to annual screening and risks are reduced correspondingly if a 2-year screening interval is used instead. In a recent survey of 22 screening units offering familial screening, about half were screening annually and half 2-yearly (R D MacMillan, personal communication). However, the interval cancer rate must also be taken into account.
All the remarks made here regarding women with family history are qualified by the question of their susceptibility to radiation-induced breast cancer, as discussed in an earlier section.
This paper does not wish to imply that the radiological risk is the only or even the principal factor to consider when drawing up screening protocols. But it is a factor that should be considered, and the outline of its magnitude presented here may assist those who design screening programmes in groups of younger women. In some groups of women, a decision whether or not to screen may be best arrived at on an individual basis.
|
Acknowledgments |
|---|
Received for publication January 29, 2001. Revision received July 11, 2001. Accepted for publication July 20, 2001.
|
References |
|---|
|
TopAbstractIntroductionNumbers of cancers detected...Cancers detected and induced...Subgroups of women receiving...ResultsDiscussionConclusionsReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  K. Faulkner Ethical concerns arising from screening procedures such as mammography and self-referral Radiat Prot Dosimetry, July 1, 2009; 135(2): 90 - 94. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. Berrington de Gonzalez, C. D. Berg, K. Visvanathan, and M. Robson Estimated Risk of Radiation-Induced Breast Cancer From Mammographic Screening for Young BRCA Mutation Carriers J Natl Cancer Inst, February 4, 2009; 101(3): 205 - 209. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. Faulkner, M.G Wallis, F Neilson, and C.J. Whitaker EVALUATION OF THE POPULATION DOSE TO THE UK POPULATION FROM THE NATIONAL HEALTH SERVICE BREAST SCREENING PROGRAMME Radiat Prot Dosimetry, May 14, 2008; (2008) ncn081v1. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E L Nickoloff and P O Alderson A comparative study of thoracic radiation doses from 64-slice cardiac CT Br. J. Radiol., July 1, 2007; 80(955): 537 - 544. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. L. Smathers, J. M. Boone, L. J. Lee, E. A. Berns, R. A. Miller, and A. M. Wright Radiation Dose Reduction for Augmentation Mammography Am. J. Roentgenol., May 1, 2007; 188(5): 1414 - 1421. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J Law, K Faulkner, and K C Young Risk factors for induction of breast cancer by X-rays and their implications for breast screening Br. J. Radiol., April 1, 2007; 80(952): 261 - 266. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Mascalchi, G. Belli, M. Zappa, G. Picozzi, M. Falchini, R. D. Nave, G. Allescia, A. Masi, A. L. Pegna, N. Villari, et al. Risk-benefit analysis of X-ray exposure associated with lung cancer screening in the Italung-CT trial. Am. J. Roentgenol., August 1, 2006; 187(2): 421 - 429. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. Allen, T. Demetriades, and D. J. Brenner Radiation risk overestimated. Radiology, August 1, 2006; 240(2): 613 - 614. [Full Text] [PDF]
|
|
|
|
 |
|
 J. Barentsz, S. Takahashi, W. Oyen, R. Mus, P. De Mulder, R. Reznek, M. Oudkerk, and W. Mali Commonly Used Imaging Techniques for Diagnosis and Staging J. Clin. Oncol., July 10, 2006; 24(20): 3234 - 3244. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C J Whitaker, C M Kelly, K Faulkner, and E C Stamp Influence of menopausal status and use of hormone replacement therapy on radiation dose from mammography in routine breast screening. Br. J. Radiol., July 1, 2006; 79(943): 597 - 602. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J Law and K Faulkner Radiation benefit and risk at the assessment stage of the UK Breast Screening Programme. Br. J. Radiol., June 1, 2006; 79(942): 479 - 482. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. E. Jacobi, N. J.D. Nagelkerke, J. C. van Houwelingen, and G. H. de Bock Breast Cancer Screening, Outside the Population-Screening Program, of Women from Breast Cancer Families without Proven BRCA1/BRCA2 Mutations: a Simulation Study. Cancer Epidemiol. Biomarkers Prev., March 1, 2006; 15(3): 429 - 436. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G J Heyes, A J Mill, and M W Charles Enhanced biological effectiveness of low energy X-rays and implications for the UK breast screening programme. Br. J. Radiol., March 1, 2006; 79(939): 195 - 200. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. Faulkner and K. Bennison An assessment of digital stereotaxis in the National Health Service Breast Screening Programme Radiat Prot Dosimetry, December 1, 2005; 117(1-3): 327 - 329. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. Faulkner and J. Law Mammographic breast cancer screening for women previously treated with high breast doses for diseases such as Hodgkin's Radiat Prot Dosimetry, December 1, 2005; 117(1-3): 330 - 333. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. Smans, H. Bosmans, M. Xiao, A. K. Carton, and G. Marchal Towards a proposition of a diagnostic (dose) reference level for mammographic acquisitions in breast screening measurements in Belgium Radiat Prot Dosimetry, December 1, 2005; 117(1-3): 321 - 326. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H M Warren-Forward and L Duggan Towards in vivo TLD dosimetry in mammography Br. J. Radiol., May 1, 2004; 77(917): 426 - 432. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J Law and K Faulkner Two-view screening and extending the age range: the balance of benefit and risk Br. J. Radiol., November 1, 2002; 75(899): 889 - 894. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| BJR | DMFR | IMAGING | ALL BIR JOURNALS |
