Influence of tumour stage at breast cancer detection on survival in modern times: population based study in 173 797 patients
BMJ 2015; 351 doi: https://doi.org/10.1136/bmj.h4901 (Published 06 October 2015) Cite this as: BMJ 2015;351:h4901
All rapid responses
Rapid responses are electronic comments to the editor. They enable our users to debate issues raised in articles published on bmj.com. A rapid response is first posted online. If you need the URL (web address) of an individual response, simply click on the response headline and copy the URL from the browser window. A proportion of responses will, after editing, be published online and in the print journal as letters, which are indexed in PubMed. Rapid responses are not indexed in PubMed and they are not journal articles. The BMJ reserves the right to remove responses which are being wilfully misrepresented as published articles or when it is brought to our attention that a response spreads misinformation.
From March 2022, the word limit for rapid responses will be 600 words not including references and author details. We will no longer post responses that exceed this limit.
The word limit for letters selected from posted responses remains 300 words.
It is perfectly clear from the paper by Tilanus-Lindhorst and colleaugues that breast cancer screening was available over the entire study period [1]. It is likewise very clear from my rapid response (see below) that this was a premise for my criticism, i.e. I criticize that they did not mention the change to a more sensitive screening modality over their study period. Their suggestion that a misunderstanding on my behalf is the explanation for our disagreement is therefore not correct [1].
The reason we disagree whether their conclusion that “early detection is vital” were justified is that Tilanus-Lindhorst and colleagues do not recognize a major logical fallacy in their argument. This fallacy makes them draw a very strong conclusion that cannot be substantiated by the data they present. As I have already argued (below), it is not possible to conclude that “early detection is vital” from the observation that prognosis deteriorates with later stage at detection. Indeed, this mechanism of effect for early detection is a major topic for debate in breast cancer.
Tilanius-Lindhorst and colleagues also argue that their study did not evaluate screening mammography and that my criticism of their conclusion is therefore unjustified. There are arguably four relevant tools for early detection of breast cancer, which they conclude is “vital”. As I have explained (below), increased disease awareness led to much earlier detection of breast cancer in Denmark [2], but this had no effect on breast cancer mortality which kept increasing, so this cannot be what Tilanius-Lindhorst and colleagues refer to. Regular breast self-examination is generally recommended against, as it doubles biopsy rates, with no benefit [3]. Regular clinical breast examination has just been recommended against by the American Cancer Society [4]. The only likely candidate intervention for "vital" early detection thus seems to be the Dutch breast screening programme, especially as the continued relevance of such programmes have been questioned due to improved treatment, a key component of their study question [1]. If they were in fact referring to another means to early detection of breast cancer, I am curious to hear what they were alluding to.
Tilanius-Linhorst and colleagues have not responded to my concerns. They still appear to draw conclusions far beyond what their study can justify.
1. Saadatmand S, Bretveld R, Siesling S, et al. Influence of tumour stage at breast cancer detection on survival in modern times: population based study in 173 797 patients. BMJ. 2015;351.
2. Rostgaard K, Vaeth M, Rootzén H, Lynge E. Why did the breast cancer lymph node status distribution improve in Denmark in the pre-mammography screening period of 1978-1994? Acta Oncol 2010;49:313-21.
3. Kösters JP, Gøtzsche PC. Regular self-examination or clinical examination for early detection of breast cancer. Cochrane Database of Systematic Reviews 2003, Issue 2. Art. No.: CD003373. DOI: 10.1002/14651858.CD003373.
4. Oeffinger KC, Fontham ETH, Etzioni R, et al. Breast cancer screening for women at average risk. JAMA 2015;314:1599-1614.
Competing interests: No competing interests
We thank Madeleine Tilanus-Linthorst and her co-authors for presenting further analysis from their study. Their new Table 1, appears to show that <10 % of deaths might be reasonably attributed to breast cancer at any age, which supports the case that many of the breast cancer cases (particularly the early stage disease) may be over-diagnosed.
The new Table 2 allows a crude estimate of effects of screening: although there are likely to be screen detected cases in both groups, there should be a higher proportion of screen-detected cases in the group invited to undergo population screening (women aged 55-75 years). The proportion of the total cancer cases that are T1a/b is much higher for that group (25%) compared to women not invited to undergo population screening (11%). As shown in Table 2 of the original paper(1) , women with stage T1a/b disease had the same 5 year survival as women without breast cancer matched by age and year of diagnosis. Taken together, these data strongly suggest over-diagnosis associated with mammography screening.
We agree with the authors that their "study is not an evaluation of screening". As such, it does not provide evidence to support statements such as those in the accompanying editorial(2) that “early detection [i.e. screening] is still vital in improving outcomes for breast cancer.”
1. Saadatmand S, Bretveld R, Siesling S, et al. Influence of tumour stage at breast cancer detection on survival in modern times: population based study in 173 797 patients. BMJ. 2015;351.
2. Vaz-Luis I, Burstein HJ. With better adjuvant therapy, does breast cancer stage still matter? BMJ. 2015;351.
Competing interests: No competing interests
Dr. Bell asks us to pay more attention to “overdiagnosis” which they define as ”detection of cancers which would have caused no health problems had they been left undetected and untreated” and to the effects of screening on our results.
Firstly when looking at our Table 2 (1) we can observe when comparing 5-yr relative with 5-yr overall survival of the 2006-2012 cohort:
Table 1--see attachment
This demonstrates that in women below age 50 breast cancer is a main cause of death, as shown also by other studies. This makes it quite difficult to regard any breast cancer, whether detected by the woman herself or by a mammogram, as certainly ”not causing a health problem”. And the same holds true for women between age 50-75 yr. where breast cancer adds considerably to the total mortality. We cannot decide whether detected tumours are “overdiagnosis” and could be left untreated apart from low-grade DCIS, for which several trials are started for “watchful waiting”. Above age 75 yr. we have no screening, but it is still worthwhile not to detect tumours late if the woman has a life expectancy of ≥ 10 yr.
We do agree with dr. Bell and co-authors, that the early detected breast cancers should not be overtreated. Women who are diagnosed with early stage breast cancer should be well aware of their excellent prognosis. We therefore are glad we could publish our nationwide results in The BMJ to inform patients and their doctors and are happy when they are widely read and well interpreted. For instance, our result that “contalateral breast cancer did not influence survival” and that ”breast conserving therapy seemed even a bit better than mastectomy for survival” (which most likely is not completely explained by comorbidity), may help not to choose “overtreatment”. Also the very beneficial effect if applied of additional hormonal and anti-HER2 therapy, but far more modest effect of adjuvant chemotherapy can be helpful when in doubt.
Dr. Bell and co-authors ask us to differentiate also in our multivariable analyses between symptomatic and screen-detected cancers.
We therefore have performed the multivariable analyses of table 4 of the clinicopathological variables for overall mortality in patients with invasive breast cancer in the Netherlands Cancer Registry 2006-2012 separately for the group that is invited for screening (50-75 yr) and the other patients (< 50 yr and > 75 yr.) Table1.
Table 2--see attachment
In both groups tumour size and nodal stage had a significant influence on survival, corrected for age, tumour biology hormone receptor and HER2 status, breast surgery, axillary lymph node dissection, chemotherapy, hormonal therapy, Her2-targeted therapy, radiotherapy, and second primary breast cancer. This is as expected as we did take tumor biology into account in the 2006-2012 BMJ multivariable analyses. That early detection is important also for prognosis of more aggressive cancers is supported by publications on the good prognosis of BRCA1 cancers (mainly grade 3, triple negative), if detected early. (2,3)
Dr. Bell also asks us to correct for lead time,” the amount of time by which the diagnosis has been advanced by screening”. Lead time correction is applicable when survival of a screened and an unscreened group are compared, which we do not in our study. Our study is not an evaluation of screening. We currently cannot on individual base correct for lead time in patients that were screen-detected, as mode of detection is being verified for completeness and correctness by the Netherlands Cancer Registry at this time. Lead time for mammography screening is usually estimated at 2 years.(4) If we assume that all T1 tumors in our study are screen-detected and all T2,T3 and T4 tumours are symptomatic (which of course is not completely true), it is clear that even if we subtract 2 years of all the T1 survival (Figure 1), T1 survival is still considerably better than ≥T2 (5-yr relative survival T1a/b 100%, T1c 98%; ,3-yr relative survival T2 94%, T3 87% in 2006-2012). This is an overcorrection.
We therefore can state that early stage at detection really improves survival both long and short-term, and does not make the period that a woman feels a patient only longer.
The transition of film-screen to digital mammography took place in the Netherlands gradually from 2006 onwards. This transition was both for the analysis of symptomatic women as in the national screening programme, but did not lead to a large shift towards lower stages or increase of DCIS. The study on which we base this conclusion is now under review for publication. We do not think therefore you can decide that the contribution of digital mammography is only overdiagnosis, which we have further discussed in our first paragraph.
Our study is not an evaluation of screening, nor does it weigh benefits and harms of different tools for early detection, it does show the importance of early detection.
1.Saadatmand S, Bretveld R, Siesling S, et al. Influence of tumour stage at breast cancer detection on survival in modern times: population based study in 173 797 patients. BMJ 2015;351.
2.Brekelmans CT, Tilanus-Linthorst MM, Seynaeve C, et al.Tumour characteristics, survival and prognostic factors of hereditary breast cancer from BRCA2-, BRCA1- and non-BRCA1/2 families as compared to sporadic breast cancer cases. Eur J Cancer 2007;43:867-76.
3.Saadatmand S, Obdeijn IM, Rutgers EJ. et al. Survival benefit in women with BRCA1 mutation or familial risk in the MRI screening study (MRISC). Int J Cancer. 2015 Oct 1;137(7):1729-38. doi: 10.1002/ijc.29534. Epub 2015 Apr 17.PMID:25820931.
4. de Gelder R, Heijnsdijk EA, van Ravesteyn NT, Fracheboud J, Draisma G, de Koning HJ. Interpreting overdiagnosis estimates in population-based mammography screening. Epidemiol Rev 2011;33:111-21.
Competing interests: No competing interests
Clinical Breast Examinations (CBEs) were proved very sensitive in detecting tumours, even smaller than 1 cm, even those that mammography scanning missed in certain age groups, especially when trained medical doctors perform them. [1][2][3]
You asked for supporting reference studies of my statement that "all breast tumours of 1 cm or more can be detected by a trained GP, midwife, or Gynecologist during routine palpatory examinations", so I provide them.
Overzealous frequent aggressive multimodal device scanning of women's breasts in order to detect minuscule tumours of <1cm, arbitrarily claiming early findings and consequent early surgery/radiotherapy/chemotherapy increases patient survival, is proved wrong by this excellent research study of yours.
References
[1] Campbell HS, Fletcher SW, Pilgrim CA, Morgan TM, Lin S. Improving physicians’ and nurses’ clinical breast examination: a randomized controlled trial. Am J Prev Med. 1991;7:1–8.
http://www.ncbi.nlm.nih.gov/pubmed/1867894
[2] Baines CJ, Miller AB. Mammography versus clinical examination of the breasts. J Natl Cancer Inst Monogr. 1997;(22):125–9.
http://www.ncbi.nlm.nih.gov/pubmed/9709288
[3] Mary B. Barton, MD, MPP; Russell Harris, MD, MPH; Suzanne W. Fletcher, MD, MSc
JAMA. 1999;282(13):1270-1280.
http://jama.jamanetwork.com/article.aspx?articleid=191969
Competing interests: No competing interests
Authors' Reply
Dr. KJ Jørgensen agrees with us, that early breast cancer stage at detection is very important for prognosis.
Our main study question is, whether this still holds true in modern times, in which over half of the patients get apart from primary therapy with surgery (and radiotherapy if breast conserving) additional systemic therapy in order to prevent metastasis. And, if stage is still important, which difference in stage is clinically relevant. (ref 1)
All T1c tumours ( 1-2 cm) detected in the Netherlands between 2006-2012 together had 98% relative survival at 5-years, T2 tumours 92% and T3 tumours 81% etc. (table 2). And taking tumor biology into account and all the given therapies, both primary and adjuvant, tumour stage at detection had a major influence on survival. (Table 4). Tumour stage had also major influence on prognosis in the cohort with long-term follow-up at multivariable analysis. (Table 3)
These analyses together make the conclusion that early detection is vital also in modern times, with the current effective therapies, certainly justified. If all (nearly all) metastases could be prevented by adjuvant therapy, stage would have no influence on survival anymore. Therefore our research question was also justified.
We get the impression from the remarks: “overdiagnosis the most serious harm of the intervention they investigate” (4th paragraph) and “no data of fewer late stage cancer in the latter period”, that dr. Jørgensen has read our study as a comparison of two periods one with and one without screening. But as stated in the second paragraph of the introduction, screening was already fully implemented in the Netherlands for all women between age 50-75 from 1998 on, also in our older 1999-2005 cohort therefore.
We do not investigate the intervention “screening” in this study.
All the flaws dr. Jørgensen contributes to our study seem to come from this misinterpretation. Our conclusion that “detecting cancers early is of vital importance” does not say what is the best tool to this aim, nor does it weigh the balance between harm and benefits of different methods.
In other studies it has been shown that for instance the breast cancers in BRCA1 carriers grow at any age twice as fast as in the general population.(2) This requires adjustments to the tools for early detection. But as several studies have shown, if detected early also tumours with this aggressive biology (high grade, triple negative) have good prognosis.(3, 4) This is in accordance with our findings.
Our result that survival of all tumours ≤ 1cm have a 5 yr. relative survival of 100% cannot be interpreted as a sign of overdiagnosis and serious harm of an intervention, as tumours were detected ≤ 1cm in all age groups.
Reference
1. Saadatmand S, Bretveld R, Siesling S, Tilanus-Linthorst MMA.Influence of tumour stage at breast cancer detection on survival in modern times: population based study in 173 797 patients. BMJ 2015;351:h4901.
2. Tilanus-Linthorst MM, Obdeijn IM, Hop WC, et al. BRCA1 mutation and young age predict fast breast cancer growth in the Dutch, United Kingdom, and Canadian magnetic resonance imaging screening trials. Clin Cancer Res 2007;13:7357-62.
3. Saadatmand S, Obdeijn IM, Rutgers EJ. et al. Survival benefit in women with BRCA1 mutation or familial risk in the MRI screening study (MRISC). Int J Cancer. 2015 Oct 1;137(7):1729-38. doi: 10.1002/ijc.29534. Epub 2015 Apr 17.PMID:25820931
4. Brekelmans CT, Tilanus-Linthorst MM, Seynaeve C, et al.Tumour characteristics, survival and prognostic factors of hereditary breast cancer from BRCA2-, BRCA1- and non-BRCA1/2 families as compared to sporadic breast cancer cases. Eur J Cancer 2007;43:867-76.
Response by the authors on the second comment.
In the second comment it is stated that “it is useless to detect <1 cm tumours by aggressive radiological methods”. We indeed revealed an equal survival in T1a (≤ 5 mm.) and T1b (6-10 mm) tumours, which could plead for a change in grouping/staging the tumours. We did however already see an increase in mortality in the T1c (11-20 mm.) tumours. We therefore think that the detection of breast cancers ≤ 1 cm certainly should not be discouraged. Our study does not say what is the best tool to this aim, nor does it weigh the balance between harm and benefits of different methods. We cannot say anything about false positive, because we did not study “screening”, or certain screening methods.
We cannot support the statements made by the commentator “that all tumors < 1 cm” (or was meant > 1 cm?) “will be detected by a trained GP, midwife, or Gynecologist during routine palpatory examinations”. We do not know any study supporting this. In the Dutch screening setting over half of the cancers are not palpable, even knowing from the imaging where in the breast they are, and among them are certainly tumours > 1 cm. But the author may be right that in a setting where for instance mammography is not available, breast examination can be used as a tool to detect breast cancers earlier.
Competing interests: No competing interests
This study demonstrates that tumor size and stage at detection is important to prognosis. This is a well-established truth that does not need further confirmation and any other result would be very surprising. It is true for most solid cancers, but does not mean that attempts at earlier detection are “vital” or even effective at all. As no data is presented to demonstrate that there are fewer late-stage cancers in the latter period of observation, this claim from the authors is unfounded. The paper does not show rates per 100.000, only percentual distributions of stages. This is a serious flaw when increasing overdiagnosis over time is likely, as overdiagnosis will produce the presented findings, even when screening has no benefit. Using 5 year survival in this setting is also a serious flaw, as survival time from diagnosis will also increase with more overdiagnosis, even when screening is entirely ineffective.
Tumor stage at diagnosis may be determined largely by the underlying biology of the tumor, which may not be amenable to attempts to earlier detection. This is the reason we need randomized trials of cancer screening, and that some cancer screening programmes do not work (i.e. neuroblastoma and ovarian cancer screening). In Denmark, the average size of breast tumors declined one full centimeter from the late 1970’sto 1980’s, well before any screening [1]. This had no effect on breast cancer mortality, which kept increasing, although this is twice the decline in average size seen in the randomized trials of breast screening. Thus, even if the authors had demonstrated a true stage shift, and not just a change in percentage distribution, this would not show that screening was effective, or demonstrate that earlier detection is “vital” – this can happen independently of screening and without any effect on mortality.
The authors need to publish an Erratum, denouncing this part of their conclusion. Especially as they announce in their paper that the findings will be widely published on the home pages of political advocacy groups (i.e. patient organisations such as cancer charities).
Their conclusion that, since survival has increased for all stages, treatment has improved, is more valid. But we knew that already from the recently published overview by the EBTCG in Lancet this year [2], and the type of design used here is susceptible to so many confounders that determining the contribution of each factor; screening and improved treatment respectively, is not possible.
The authors do not mention how factors such as the introduction of digital mammography over their observation period could have increased the proportion of small tumors due to increased overdiagnosis. This is another serious flaw. In fact, they do not mention anywhere that many of the tumors they include are overdiagnosed. They do not mention overdiagnosis at all, although this is the most serious harm of the intervention they investigate and a major factor contributing to the observations they make. For example, overdiagnosis influence their finding that the relative survival of DCIS was 101% in the latter observation period (likely also due to selection bias - healthy attendees who are then overdiagnosed, which is not mentioned). That breast cancer mortality has decreased 50% in women below the screening age, but only 35% in women in screened age groups [3], is testimony to the major influence of improved therapy, but not early detection. Such a simple comparison is also lacking here. In stead, the authors seem to lump all age groups. Overdiagnosis is very likely also an important contributing factor to their finding that the 5-year survival rate for tumors below 1 cm at detection was 100%. This may not be such a good thing as it appears in the paper as this, to a substantial extent, is a result of the major harm of the intervention. Screen-detected small tumors that would never have progressed invariably has 100% 5-year survival.
1. Rostgaard K, Vaeth M, Rootzén H, Lynge E. Why did the breast cancer lymph node status distribution improve in Denmark in the pre-mammography screening period of 1978-1994? Acta Oncol 2010;49:313-21.
2. Early Breast Cancer Trialists' Collaborative Group (EBCTCG). Aromatase inhibitors versus tamoxifen in early breast cancer: patient-level meta-analysis of the randomised trials. Lancet 2015;386:1341-52.
3. Autier P, Boniol M, LaVecchia C, Vatten L, Gavin A, Héry C, et al. Disparities in breast cancer mortality trends between 30 European countries: retrospective trend analysis of WHO mortality database. BMJ 2010;341:c3620.
Competing interests: No competing interests
Aggressive radiological methods for detecting breast tumours <1cm are useless, and could increase false positives, this study shows.
Even invasive breast cancers of 1cm size eventually demonstrate equal survival rates with smaller cancers, at 100% in 15 years.
It is difficult for a trained GP, midwife, or Gynecologist to miss breast tumours of 1 cm during routine palpatory examinations.
Results from this study support the case against frequent multimodal device scanning of women's breasts in order to detect minuscule tumours.
Competing interests: No competing interests
The need to account for Screening and Over-diagnosis
Katy Bell, Gemma Jacklyn, Andrew Hayen, Les Irwig and Alexandra Barratt
Saadatmand and colleagues1 demonstrate a substantial increase in the proportion of women having effective treatment between 1999-2005 and 2006-2012, many of whom may have had their lives extended. We are concerned by the authors’ statements on the effects of early cancer detection and the inferences made in the accompanying editorial2 that this provides “a powerful albeit indirect argument in favour of screening mammography”. Our concerns center on the apparent lack of consideration given to the effects of screening and over-diagnosis of breast cancer3.
The authors need to allow for the method of detection (screening vs clinical presentation) in the analysis of the effects of prognostic factors on overall mortality for each cohort. They report that “tumour size and nodal status still had a significant and major influence”. But as screening is associated with smaller tumour size and lower stage at diagnosis (lead time bias) as well as with survival independent of this (because of length time bias and over-diagnosis)4, it is likely to be confounding the associations between tumour size/stage and survival. It would be interesting to see the results for Tables 3 and 4 when the method of detection is included as a covariate in the model, or when models are stratified by the method of detection. These additional analyses are likely to only adjust partially for the biases introduced by screening, including over-diagnosis. For example, asymptomatic breast cancer which is diagnosed outside of the population screening program (e.g. opportunistic mammograms) may not be identified as being screen detected. But they may at least give some indication of the inflation of survival estimates due to screening.
Overall the study findings are suggestive of increasing over-diagnosis (detection of cancers which would have caused no health problems had they been left undetected and untreated). For example, cancers diagnosed in the later period were smaller, more often lymph node negative and more often low grade than cancers diagnosed in the earlier period. The second cohort (women diagnosed with breast cancer 2005-2012) coincides with the transition from screen-film to digital screening mammography in Netherlands. Digital screening has a higher overall cancer detection rate but similar interval cancer rate compared to screen-film mammography5, and may be associated with higher rates of over-diagnosis. In addition, there appears to be progressively more detection of early stage cancer (in situ and Stage 1) for women aged ≥70 years during the time period of the current study6. This may reflect increased screening in this age group both within the population screening program and opportunistic screening outside of it (for women >75 years). These changes in the screening technology and in the proportion of older women being screened, may account for some of the differences in relative survival found between cohorts in the present study, with a higher proportion of cancers over-diagnosed – and over-treated - in the more recent cohort. Separate presentation of the results in Table 2 for screen-detected and clinically-detected cancers would be helpful in deciding the extent to which changes in screening may explain the difference in outcomes between the two cohorts.
By providing evidence on the extent to which biases associated with screening, including over-diagnosis, explain the improved survival in more recent years, readers will be better able to decide whether they agree with the authors’ conclusion that “Diagnosis of breast cancer at an early tumour stage remains vital”.
1. Saadatmand S, Bretveld R, Siesling S, et al. Influence of tumour stage at breast cancer detection on survival in modern times: population based study in 173 797 patients. BMJ 2015;351.
2. Vaz-Luis I, Burstein HJ. With better adjuvant therapy, does breast cancer stage still matter? BMJ 2015;351.
3. Barratt A. Overdiagnosis in mammography screening: a 45 year journey from shadowy idea to acknowledged reality. BMJ 2015;350.
4. Bell K, Barratt A, Hayen A. When talking about cancer screening, survival rates mislead. 2015. http://theconversation.com/when-talking-about-cancer-screening-survival-....
5. Nederend J, Duijm LE, Louwman MW, et al. Impact of the transition from screen-film to digital screening mammography on interval cancer characteristics and treatment - a population based study from the Netherlands. Eur J Cancer 2014;50(1):31-9.
6. de Glas NA, de Craen AJM, Bastiaannet E, et al. Effect of implementation of the mass breast cancer screening programme in older women in the Netherlands: population based study. BMJ 2014;349.
Competing interests: No competing interests
Re: Influence of tumour stage at breast cancer detection on survival in modern times: population based study in 173 797 patients
Similar conclusions from another large study:
"The mortality rate in 15 years was similar for breast tumours 1 mm in diameter (7.3%) and tumours 9 mm in diameter (7%), despite a 763.4-fold difference in volume (0.5–381.7 mm3)."
"Among all 819,647 patients with invasive breast cancer, the rate of increase in lymph node metastasis peaks for tumours 10 mm in size, after which the rate decreases.
This implies that the propensity to metastasize decreases as the tumour grows beyond 1 cm."
Reference
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6022519/
Competing interests: No competing interests