In the last two decades, breast cancer survival has improved. Reduced mortality from this disease has been attributed to a combination of widespread mammographic screening, the introduction of effective systemic treatment modalities, and progress in radiotherapy and surgery (Berry et al., 2005; Benson et al., 2009; Autier et al., 2010). However, the contribution of screening to this reduction is unclear, as outcomes may be affected by several biases such as lead-time bias (screen-detected cancers are detected earlier in their natural course than those found outside of screening), length bias (screening tends to detect slow-growing tumors, which spend longer time in the asymptomatic phase), and selection bias (the screened population is not representative of the general population).
Previous works that have focused on breast cancer survival and detection mode have reported that screen detection confers an additional survival benefit beyond the stage shift (Shen et al., 2005; Wishart et al., 2008) and reduces the risk of systemic recurrence compared with symptomatic cancers at a similar stage (Joensuu et al., 2004). However, part of this benefit remains unexplained as it is difficult to obtain molecular information from a large series of tumors for inclusion in statistical models (Vitak et al., 1997; Immonen-Räihä et al., 2005; Shen et al., 2005; Bordas et al., 2007; Zackrisson et al., 2007; Dawson et al., 2009; Lawrence et al., 2009).
It is already known that tumors detected during screening are related to clinical–pathological features with a better prognosis, such as low grade or hormone-receptor expression, compared with those detected by other means (i.e. outside of screening or as interval cancers; Bordas et al., 2007). Wider differences have been described between screen-detected and interval cancers (tumors that manifest clinically between a normal screening result and the following invitation for screening), which are associated with a delay in diagnosis that could potentially worsen prognosis and consequently impair the program’s efficiency. Nevertheless, the greatest differences have been found between screen-detected cancers and true interval cancers (tumors with a short preclinical phase and no suspicious finding after radiological review of the previous mammogram). Among this subgroup, increased tumor cell proliferation (Vitak et al., 1997) and a higher occurrence of triple-negative tumors [tumors that lack expression of estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER2)] have been reported (Domingo et al., 2010; Rayson et al., 2011). To date, this biomarker pattern lacks the benefit of specific adjuvant therapy and is associated with a poor prognosis (Chacón and Costanzo, 2010).
Although differences at diagnosis have been reported between screen-detected and true interval cancers, only two studies (Vitak et al., 1997; Rayson et al., 2011) have taken this latter group into consideration when evaluating clinical outcomes, and both have reported significantly poorer survival. However, neither of these studies included information on phenotype and a group of women who did not participate in the screening.
The aim of the present study was to evaluate the role of the diagnostic method and biological characteristics in relapse-free survival and overall survival in breast cancer patients.
The study was carried out among women diagnosed with breast cancer in a reference hospital in the city of Barcelona (Spain) between 1995 and 2008. This center is publicly funded, serves an area with around 300 000 inhabitants, and has run a population-based breast cancer screening program since 1995.
The screening program’s target population (∼90 000 women) is women aged 50–69 years, and involves women who are invited to undergo mammography every 2 years. Mediolateral, oblique, and craniocaudal views are available for each breast. All mammograms are read by two radiologists, and when double reading leads to different assessments, a third radiologist serves as a tiebreaker. The program is based on the European Guidelines for Quality Assurance in Mammographic Screening (Perry et al., 2006) and its results meet the Europe Against Cancer standards, although the participation rate barely reached the standard level of 70%.
This study was approved by the Ethics Committee and informed consent was obtained to supply tumor biopsy material for pathologic evaluation.
A total of 1432 patients with breast cancer aged 50–69 years were identified from the hospital-based cancer registry during the study period. Of these cancer cases studied, 740 were detected in women attending the breast cancer screening program (screen-detected cancers), 98 emerged as interval cancers between screening mammograms, and 594 cancers were diagnosed among women who did not participate in the screening program (symptom-detected cancers) and who were referred with breast abnormalities, typically palpable lesions, by their primary care physicians or were self-referred to the hospital.
Interval cancers were retrieved by merging the mammography register with the hospital-based cancer registry; in this process, we identified 98 interval cancers diagnosed and treated in our setting. For these, we aimed to perform a radiological review of both screening and diagnostic mammograms to classify them into the five subtypes, following European Guideline recommendations (Perry et al., 2006). Both mammograms were available in 80 cases, and two experienced radiologists performed the retrospective review. Screening mammograms were first reviewed independently, and in a second review, the radiologists assessed screening and diagnostic mammograms with the histological information to determine whether any abnormalities detected on the screening films corresponded to the site of the subsequent interval cancer. Interval breast cancers were definitively classified into five subtypes: true interval cancer (n=34), false negative (n=13), occult tumors (n=14), minimal signs (n=4), and unclassifiable (n=15). A complete description of the interval cancer identification process has been published previously (Domingo et al., 2010).
Because of difficulties in obtaining retrospective information on such a large number of cases, a random sample of screen-detected (n=97) and symptom-detected cancers (n=97) was selected. These two samples were compared with all true interval cancers (n=34). In-situ cancers and patients with stage IV breast cancer at diagnosis were excluded from the analyses.
Tumor-related data [pathological tumor–node–metastasis (TNM) status, histological type, histological grade], patient-related data (age), and vital status at the end of follow-up were obtained from the hospital-based cancer registry. Data on treatment, recurrences, and immunohistochemical information (ER, PR, HER, and p53 status) were obtained from the registry of pathology, if available, and from review of clinical records. Further immunohistochemical analyses were carried out in tumor samples in which biomarker expression had not been determined previously. Information on adjuvant treatment (radiotherapy, chemotherapy, hormonotherapy, and targeted therapies) and follow-up was obtained from the clinical records.
ER, PR, p53, and HER2 are routinely determined during the diagnostic process in our hospital. Samples lacking information on biomarker expression were analyzed following the same procedures as those used in clinical practice. Immunohistochemical staining was performed on paraffin block sections with the tissue specimens fixed in 10% neutral-buffered formalin for 24 h [ER clone ID5, 1 : 50 (Dako, Glostrup, Denmark); PR clone PgR636, 1 : 200 (Dako); p53 clone DO7, 1 : 50 (Novocastra Laboratories, Newcastle, UK); for Her2/neu protein overexpression (HercepTest; Dako)]. In accordance with standard guidelines, positivity for ER, PR, and p53 was based on more than 10% of the cells testing positive. For HER2, scores of 0–1 in the HerceptTest kit were considered negative, whereas a HerceptTest score of 3 was considered positive. Equivocal scores of 2 were confirmed by fluorescent in-situ hybridization as positive when HER2/neu oncogene amplification was detected.
On the basis of the expression of ER, PR, and HER2, the tumors were classified into four phenotypes: (a) luminal A (ER+, PR+, HER−); (b) luminal B (ER or PR+, HER− or ER+, PR+/−, HER+); (c) HER2 (ER−, PR−, HER2+); and (d) triple negative (ER−, PR−, HER−) (Perou et al., 2000; Cheang et al., 2009).
Follow-up of cancer cases
Locoregional recurrence was defined as disease recurrence within the ipsilateral breast or chest wall, in the ipsilateral axillary nodes, internal mammary nodes, or supraclavicular nodes. Distant recurrence was defined as disease recurrence in sites other than the breast or regional lymph nodes (bone, skin, or visceral metastasis).
Disease-free survival was defined as the time from diagnosis to the first occurrence of one or more of the following: a local or regional recurrence, cancer in the contralateral breast, distant metastasis, and second primary carcinoma, whichever occurred first. Overall survival was defined from the date of diagnosis to death from any cause.
When disease-free survival was computed, women lost to follow-up or those who died were censored either at last visit or at death. For overall survival, patients were censored at the date of their last hospital visit. The median follow-up period was 5.13 years.
Contingency tables were calculated to compare possible differences in patient and histopathological characteristics among the study groups. Statistical significance was assessed using χ2-tests. Survival curves were generated using the Kaplan–Meier method and were compared by the log-rank test. Kaplan–Meier estimates of 5-year disease-free and overall survival rates after diagnosis were computed with 95% confidence intervals (CI).
Cox proportional hazard regression analyses were carried out to evaluate survival differences between screen-detected, true interval, and symptom-detected cancers, controlling for known prognostic and predictive factors such as age, TNM stage, and phenotype in an attempt to control lead time and length biases related to screening. Unadjusted and adjusted hazard ratios and 95% CI were computed and are shown for our main variable of interest (detection method). We computed a baseline regression model that included only the detection method, and gradually, all other study variables were added to control for their potential effect on survival times. The proportional hazards assumption was ascertained by assessment of log−log survival plots. All calculations were carried out using the statistical software SPSS, version 12.0 (SPSS Inc., Chicago, Illinois, USA) and R, version 2.12.2 (R Development Core Team, 2011). All P-values were two-sided and values less than 0.05 were considered statistically significant.
The analyses included 228 patients with breast cancer, of which 97 were screen-detected, 34 were true interval cancers, and 97 were cancers detected symptomatically outside of the screening program. The clinical and pathological characteristics for the three detection groups are shown in Table 1. At diagnosis, clinically detected tumors (symptom-detected and true interval cancers) were at more advanced stages, larger, more frequently classified as lymph node positive, and poorly differentiated compared with screen-detected cancers. The most frequent phenotype in all three groups was luminal A, but the highest percentage was observed among screen-detected cancers (66.3%). However, the most frequent occurrence of the triple-negative phenotype was found among true interval cancers (28.1%), in contrast to the proportions observed among screen-detected and symptom-detected cancers (3.5 and 10.7%, respectively; P=0.002).
The median follow-up was 6.7 years (range, 0.1–14.0) for women with screen-detected cancer, 3.9 years (range, 0.5–11.7) for women with true interval cancer, and 6.2 years (range, 0.5–15.2) for symptomatic women. Figure 1 shows the disease-free survival curves by detection mode and molecular phenotypes. Figure 2 shows the overall survival curves by these same factors. Disease-free survival and overall survival were worse in true interval cancers than in screen-detected and symptomatic cancers. Disease-free survival was longer in luminal cancers (both A and B) than in HER2 and triple-negative tumors (log-rank test=12.1; P=0.007). Kaplan–Meier estimates of the 5-year disease-free survival rates after diagnosis for screen-detected, true interval, and symptom-detected cancers were 87.5% (95% CI, 80.5–95.2%), 64.1% (95% CI, 46.4–88.5%), and 79.4% (95% CI, 71.0–88.8%), respectively. Kaplan–Meier estimates of the 5-year overall survival rates after diagnosis for screen-detected, true interval, and symptom-detected cancers were 94.5% (95% CI, 89.3–99.9%), 65.5% (95% CI, 47.1–91.2%), and 85.6% (95% CI, 78.3–93.6%), respectively.
Estimates of the Cox regression models to evaluate the determinants associated with relapse-free survival and overall survival are presented in Tables 2 and 3, respectively. The unadjusted model showed that true interval cancers and symptom-detected cancers had a higher risk of relapse than screen-detected cancers, the estimated hazard ratios being statistically significant for true interval cancers. However, in the adjusted model, the prognostic effect of the detection method was attenuated, especially when biological factors were included in the multivariate model. The hazard risk of death was 5.02 (95% CI, 1.87–13.48) times higher among true interval cancers, and was 2.78 (95% CI, 1.24–6.24) times higher among symptom-detected cancers, compared with screen-detected cancers. Adjustment for TNM stage attenuated both values, which nevertheless remained statistically significant. Finally, in the model including phenotype, only true interval cancers were associated with a worse prognosis.
Our results suggest that tumors detected clinically, especially true interval cancers, encompass a subgroup of tumors with different biological characteristics leading to poorer prognosis and worse outcomes than in screen-detected cancers. When adjustment for age, TNM stage, and phenotype was carried out, the detection mode remained as an independent factor for overall survival.
As expected, the pathological characteristics of screen-detected tumors were related to a better prognosis. At diagnosis, screen-detected cancers were smaller, more frequently lymph node-negative, and of lower grade than clinically detected tumors. These findings are in agreement with several publications (Vitak et al., 1997; Joensuu et al., 2004; Collett et al., 2005; Shen et al., 2005; Bordas et al., 2007; Pálka et al., 2008; Chiarelli et al., 2012). Among clinically detected tumors, 18.8% of true interval cancers were larger than 50 mm, whereas this percentage was 11.1% for symptom-detected cancers. This finding is especially important, given that true interval cancers have a short preclinical phase (sojourn time) and truly arise in less than 2 years. ER and PR positivity was more frequent in screen-detected than in clinically detected cancers, as observed in other studies (Sihto et al., 2008; Dawson et al., 2009; Mook et al., 2011; Nagtegaal et al., 2011), but not in all (Joensuu et al., 2004). In agreement with previous series (Domingo et al., 2010; Van der Vegt et al., 2010; Rayson et al., 2011), we found the highest percentage of triple-negative tumors among true interval cancers, with substantial differences when compared with both screen-detected and symptom-detected cancers.
These results, together with the findings of other researchers that relate true interval cancers with tumors with high mitotic capacity and a high cell-proliferation phase (Ki-67 and S-phase fraction; Vitak et al., 1997; Crosier et al., 1999; Kirsh et al., 2011), support the hypothesis that true interval cancers constitute a subgroup of breast cancers with rapid growth and high aggressiveness that are less likely to be detected on screening mammograms. These features may determine the probability of detection by screening programs; tumors with a shorter sojourn time (which is the time taken by tumors to grow from a mammographically detectable size to a clinically detectable size) are more likely to be detected between screening intervals (Weedon-Fekjaer et al., 2005). This is in part a reflection of the length bias associated with screening practices, in agreement with the idea that tumors detected on routine screening are not simply tumors diagnosed earlier, before becoming symptomatic, but also show biological differences. However, a percentage of screen-detected cancers also show features of worse prognosis (41.1% in stages II–III, 26.3% lymph node positive, and 3.5% presenting a triple-negative phenotype) that could benefit from screening.
Several studies have shown that screen detection remains an independent prognostic factor after adjustment for disease stage (Joensuu et al., 2004; Shen et al., 2005; Wishart et al., 2008) and some biological characteristics (Gill et al., 2004; Sihto et al., 2008; Dawson et al., 2009). Most of these studies, which do not differentiate interval cancers, recommend taking the detection method into account when estimating individual prognosis (Joensuu et al., 2004; Mook et al., 2011). In agreement with previous series, our results showed better survival outcomes among tumors detected by mammography than among those detected by other means (Vitak et al., 1997; Joensuu et al., 2004; Bordas et al., 2007; Zackrisson et al., 2007). Nonetheless, most of the studies considered all interval cancers together, which may have attenuated the worse outcome in the true interval cancer subgroup. Studies revealed that true interval cancers (Vitak et al., 1997; Van der Vegt et al., 2010; Rayson et al., 2011) showed a trend toward decreased relapse-free and overall survival when compared with cancers detected by other means. However, only Vitak et al. (1997) compared true interval with both screen-detected and symptom-detected cancers, whereas the remaining studies did not include this last group.
Unadjusted Cox models for disease-free survival showed the method of detection as an independent factor for relapse prediction, consistent with previous series including true interval cancers (Vitak et al., 1997; Rayson et al., 2011). However, when tumor size and biological characteristics were included in the multivariate analyses to adjust for lead and length bias, as expected, the prognostic value of the detection mode lost its effect for recurrences. Nevertheless, for overall survival, our trends suggest that true interval cancers trigger a subgroup of breast cancers with an independent unfavorable prognostic significance beyond that explained by the conventional factors included in the present study. If confirmed, this finding may merit further investigation into its underlying biological mechanisms. Recently, some genetic and epigenetic mechanisms have been related to interval malignancies, such as the methylation process of specific genes (Suijkerbuijk et al., 2011) or the action of growth factors produced in the breast stroma in response to tumor aggressiveness (Li et al., 2005). Further efforts should be made in the future to improve understanding of the genetic mechanisms related to cancer aggressiveness, tumor cell proliferation, and the process of carcinogenesis to improve the early detection of fast-growing tumors.
This study has some limitations, the main one being the small sample size. Given the difficulties in identifying and classifying interval cancers, to date, few studies have focused on true interval cancers. However, the current study adds to previous analyses limited by incomplete pathological data or a failure to restrict the analyses to true interval cancers. Second, misclassification of the detection mode cannot be excluded. Some interval cancers could be classified as screen-detected if symptomatic women waited for the screening visit instead of making an appointment with a physician. This misclassification would attenuate the effect studied. Thus, the survival difference in favor of screened women might be greater than that observed because of the inclusion of some women with symptomatic cancers in the screened cohort. In addition, the sociodemographic characteristics of women participating in the screening program, whose interval cancer was diagnosed and treated in other hospitals, did not differ from those of women treated in our setting. However, as participation in screening practices may be affected by several selection factors, we considered a group of women not affected by screening in the analyses.
This study has some strengths. To our knowledge, this is the first work focused on true interval cancer characterization that analyzes prognosis by considering cancers detected both inside and outside screening, in addition to the molecular profile. This design allows us to control the main biases that affect the outcomes of the screening practices. Moreover, all analyses were carried out with information on phenotype and other biological markers from recently diagnosed patients (1996–2008) who received homogeneous adjuvant treatment on the basis of the oncology protocol of our institution: chemotherapy based on anthracyclines or anthracyclines and taxanes, hormone therapies (tamoxifen or aromatase inhibitors), and targeted therapies (trastuzumab has been prescribed to patients with HER2 overexpression since 2006) in the modern era.
These results suggest that screening programs detect a major proportion of tumors with more favorable biological characteristics, which could partly explain the better survival of patients with screening-detected tumors. Moreover, our results show that true interval cancers share a subset of features of worse prognosis, related both to their growth rate and short sojourn time, and to their clinical course. Further understanding of their biologic features and individual determinants could increase the benefits of screening: in the short term, by increasing the sensitivity of the programs and aiding the choice of optimal screening interval for specific subsets of women at high risk, and in the long term, by increasing the early detection of tumors with a less favorable natural history.
The authors acknowledge the contribution of the Hospital del Mar Tumor Registry (Barcelona) in providing tumor-related data. They also thank Cristina Hernández and Marta Román for their assistance in data management and Teresa Baró for her technical assistance in performing the immunohistochemical analysis.
This study was partially supported by CIBER de Epidemiología y Salud Pública (AE08_004).
Conflicts of interest
There are no conflicts of interest.
Autier P, Boniol M, La Vecchia 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;341:1–7
Benson JR, Jatoi I, Keisch M, Esteva FJ, Makris A, Jordan VC. Early breast cancer. Lancet. 2009;373:1463–1479
Berry DA, Cronin KA, Plevritis SK, Fryback DG, Clarke L, Zelen M, et al. Effect of screening and adjuvant therapy on mortality from breast cancer. N Engl J Med. 2005;353:1784–1792
Bordas P, Jonsson H, Nystrom L, Lenner P. Survival
from invasive breast cancer among interval cases in the mammography screening programmes of northern Sweden. Breast. 2007;16:47–54
Chacón RD, Costanzo MV. Triple-negative breast cancer. Breast Cancer Res. 2010;12(Suppl 2):1–9
Cheang MC, Chia SK, Voduc D, Gao D, Leung S, Snider J, et al. Ki67 index, HER2 status, and prognosis of patients with luminal B breast cancer. J Natl Cancer Inst. 2009;101:736–750
Chiarelli AM, Edwards SA, Sheppard AJ, Mirea L, Chong N, Paszat L, et al. Favourable prognostic factors of subsequent screen-detected breast cancers among women aged 50–69. Eur J Cancer Prev. 2012;23:1–8
Collett K, Stefansson IM, Eide J, Braaten A, Wang H, Eide GE, et al. A basal epithelial phenotype is more frequent in interval breast cancers compared with screen detected tumors. Cancer Epidemiol Biomarkers Prev. 2005;14:1108–1112
Crosier M, Scott D, Wilson RG, Griffiths CD, May FE, Westley BR. Differences in Ki67 and c-erbB2 expression between screen-detected and true interval breast cancers. Clin Cancer Res. 1999;5:2682–2688
Dawson SJ, Duffy SW, Blows FM, Driver KE, Provenzano E, LeQuesne J, et al. Molecular characteristics of screen-detected vs. symptomatic breast cancers and their impact on survival
. Br J Cancer. 2009;101:1338–1344
Domingo L, Sala M, Servitja S, Corominas JM, Ferrer F, Martínez J, et al. Phenotypic characterization and risk factors for interval breast cancers in a population-based breast cancer screening program in Barcelona, Spain. Cancer Causes Control. 2010;21:1155–1164
Gill PG, Farshid G, Luke CG, Roder DM. Detection by screening mammography is a powerful independent predictor of survival
in women diagnosed with breast cancer. Breast. 2004;13:15–22
Immonen-Räihä P, Kauhava L, Parvinen I, Holli K, Kronqvist P, Pylkkänen L, et al. Mammographic screening reduces risk of breast carcinoma recurrence. Cancer. 2005;103:474–482
Joensuu H, Lehtimaki T, Holli K, Elomaa L, Turpeenniemi-Hujanen T, Kataja V, et al. Risk for distant recurrence of breast cancer detected by mammography screening or other methods. JAMA. 2004;292:1064–1073
Kirsh VA, Chiarelli AM, Edwards SA, O’Malley FP, Shumak RS, Yaffe MJ, et al. Tumor characteristics associated with mammographic detection of breast cancer in the Ontario Breast Screening Program. J Natl Cancer Inst. 2011;103:942–950
Lawrence G, Wallis M, Allgood P, Nagtegaal ID, Warwick J, Cafferty FH, et al. Population estimates of survival
in women with screen-detected and symptomatic breast cancer taking account of lead time and length bias. Breast Cancer Res Treat. 2009;116:179–185
Li T, Sun L, Miller N, Nicklee T, Woo J, Hulse-Smith L, et al. The association of measured breast tissue characteristics with mammographic density and other risk factors for breast cancer. Cancer Epidemiol Biomarkers Prev. 2005;14:343–349
Mook S, van’t Veer LJ, Rutgers EJ, Ravdin PM, van de Velde AO, van Leeuwen FE, et al. Independent prognostic value of screen detection in invasive breast cancer. J Natl Cancer Inst. 2011;103:585–597
Nagtegaal ID, Allgood PC, Duffy SW, Kearins O, Sullivan EO, Tappenden N, et al. Prognosis and pathology of screen-detected carcinomas: how different are they? Cancer. 2011;117:1360–1368
Pálka I, Kelemen G, Ormandi K, Lazar G, Nyari T, Thurzo L, et al. Tumor characteristics in screen-detected and symptomatic breast cancers. Pathol Oncol Res. 2008;14:161–167
Perou CM, Sorlie T, Eisen MB, van de Rijn M, Jeffrey SS, Rees CA, et al. Molecular portraits of human breast tumours. Nature. 2000;406:747–752
Perry N, Broeders M, de Wolf C, Törnberg S, Holland R, von Karsa L European guidelines for quality assurance in breast cancer screening and diagnosis. 20064th ed. Luxembourg Office for Official Publications of the European Communities
Rayson D, Payne JI, Abdolell M, Barnes PJ, MacIntosh RF, Foley T, et al. Comparison of clinical-pathologic characteristics and outcomes of true interval and screen-detected invasive breast cancer among participants of a Canadian breast screening program: a nested case-control study. Clin Breast Cancer. 2011;11:27–32
R: A language and environment for statistical computing. ISBN 3-900051-07-0. 2011 Vienna, Austria R Foundation for Statistical Computing Available at: http://www.R-project.org/
Shen Y, Yang Y, Inoue LY, Munsell MF, Miller AB, Berry DA. Role of detection method in predicting breast cancer survival
: analysis of randomized screening trials. J Natl Cancer Inst. 2005;97:1195–1203
Sihto H, Lundin J, Lehtimäki T, Sarlomo-Rikala M, Bützow R, Holli K, et al. Molecular subtypes of breast cancers detected in mammography screening and outside of screening. Clin Cancer Res. 2008;14:4103–4110
Suijkerbuijk KP, van Diest PJ, van der WE. Improving early breast cancer detection: focus on methylation. Ann Oncol. 2011;22:24–29
Van der Vegt B, Wesseling J, Pijnappel RM, Dorrius MD, den Heeten GJ, de Roos MA, et al. Aggressiveness of ‘true’ interval invasive ductal carcinomas of the breast in postmenopausal women. Mod Pathol. 2010;23:629–636
Vitak B, Stal O, Manson JC, Thomas BA, Arnesson LG, Ekelund L, et al. Interval cancers and cancers in non-attenders in the Ostergotland Mammographic Screening Programme. Duration between screening and diagnosis, S-phase fraction and distant recurrence. Eur J Cancer. 1997;33:1453–1460
Weedon-Fekjaer H, Vatten LJ, Aalen OO, Lindqvist B, Tretli S. Estimating mean sojourn time and screening test sensitivity in breast cancer mammography screening: new results. J Med Screen. 2005;12:172–178
Wishart GC, Greenberg DC, Britton PD, Chou P, Brown CH, Purushotham AD, et al. Screen-detected vs. symptomatic breast cancer: Is improved survival
due to stage migration alone? Br J Cancer. 2008;98:1741–1744
Zackrisson S, Janzon L, Manjer J, Andersson I. Improved survival
rate for women with interval breast cancer — results from the breast cancer screening programme in Malmo, Sweden 1976–1999. J Med Screen. 2007;14:138–143