The identification of false negatives in a population of interval cancers: a method for audit of screening mammography

doi:10.1016/0960-9776(95)90074-8

The Breast

Volume 4, Issue 3, September 1995, Pages 183-188

https://doi.org/10.1016/0960-9776(95)90074-8 Get rights and content

Abstract

The Northern Region Breast Screening Radiology Audit Group (NRBSRAG) has developed a method of classifying interval cancers. Each member of the group independently reads a set of films which contains both interval cancers and controls from centres other than their own. A cancer is classified as ‘false negative’ only when it is correctly identified, on the previous screening film, by at least two members of the Group. The method avoids classifying a cancer as ‘false negative’, when the only evidence is based on retrospective searching for a trace of a tumour which is unlikely to be detected in any other way. This first analysis of 167 interval cancers shows: 46% ‘true intervals’, 26% ‘false negatives’, 11% ‘occult’, and 16% ‘not classifiable’.

References (15)

I. Anttinen et al.
Double reading of mammography screening films — one radiologist or two?
Clinical Radiology
(1993)
Patnick J, Muir Gray J A. Guide-lines on collection and use of breast cancer data. NHSBSP...
D.J. Goodenough et al.
Radiographic applications of receiver operating characteristic (ROC) curves
Radiology
(1974)
J.A. Hanley et al.
A method of comparing the areas under receiver operating characteristic curves derived from the same cases
Radiology
(1983)
J. Patnick
NHS Breast Screening Programme — Review
(1993)
J.R. Muhm et al.
Lung cancer detected during a screening program using 4-month chest radiographs
Radiology
(1983)

There are more references available in the full text version of this article.

Cited by (25)

Breast ultrasound in breast cancer surveillance; incremental cancers found at what cost?
2020, Breast
Citation Excerpt :
The decreased sensitivity in both mammography and US found in this study suggests under performance of these modalities in breast cancer surveillance. There is contention in regarding the classification of interval cancers as false negatives, as previous studies have indicated only a percentage of interval cancers are true false negatives [17,18]. In comparison of BIRADS and RANZCR reporting, category 3 lesions differ greatly, as BIRADS suggests a malignancy potential of <2%, probably benign in comparison to an “indeterminate” finding [19].,
To determine the diagnostic parameters of breast ultrasound (US) in the setting of routine radiological surveillance after a diagnosis of breast cancer and evaluate costs of the inclusion of breast US as well as any survival benefit of US detected cases of recurrence in surveillance.
622 patients underwent breast cancer surgery and follow up at Austin Health from July 2009 to December 2015. Retrospective data analysis was performed to determine; diagnostic parameters, financial costs of US and survival outcomes of US detected cases of recurrence.
Patients underwent 1–9 years of breast cancer surveillance, with a median of 4.24 years. 390 (62.7%) patients underwent additional breast US surveillance to mammography. 232 (38.3%) fit criteria for use of additional breast US. 199 abnormal imaging episodes occurred, leading to 16 screen detected-cases of locoregional recurrence. US alone generated 107 abnormal images and found 9 cancers. US had a sensitivity of 44.1%, specificity of 95.2% and positive predictive value of 11.7% in comparison to mammography; 20.6%, 97.4% and 9.9% respectively. US had a biopsy rate of 4.0% and lead to an incremental cancer detection rate of 0.38%. The cost of incremental cancer found was $31,463.72 AUD. Survival outcomes based on method of detection of recurrence were insignificant (p value = 0.71).
Breast US has a sensitivity of 44.1% and detected seven recurrences that were mammographically occult. Breast US has a similar PPV to mammography in surveillance. Breast US generated considerable biopsy rates and costs. Survival analysis was not able to detect any benefit of US detected cases of recurrence.
Use of previous screening mammograms to identify features indicating cases that would have a possible gain in prognosis following earlier detection
2003, European Journal of Cancer
Citation Excerpt :
In the last decade, many studies have tried to classify and describe characteristics of cancers that could have been detected at screening. Most studies focused on the review of interval cancers [3,7,9–11,13,24], but several studies also recognised the importance of rereading screen-detected or ‘incident round’ cancers, usually considered to be ‘true-positive’ [5,15–17,19–21]. In most review studies, researchers have attempted to distinguish between lesions that were overlooked or misinterpreted and lesions where the mammogram showed minimal or unusual lesion characteristics, the so-called ‘minimal signs’ [5,9,10,12,13,20,21,24].
False-negative screening mammograms generally refer to breast cancers that were overlooked or misinterpreted at screening. An important question is whether earlier detection could have made a difference in the prognosis of the women concerned. We reviewed screening and diagnostic mammograms of 234 screen-detected and interval cancer cases (aged 44–84 years) diagnosed between 1991 and 1996 in the Nijmegen breast cancer screening programme. A lesion was visible on 117 (50%) of the screening mammograms prior to the diagnosis of breast cancer. Fifty-one out of the 117 cancers had poor prognostic characteristics at diagnosis (i.e. N+ and/or T2+) and could potentially have benefited from an earlier diagnosis (‘possible gain’). The ‘possible gain’ cases were more often characterised by architectural distortion (29 vs. 10%; P=0.01) or a high-density mass (25 vs. 13%; P=0.06) on the mammogram prior to diagnosis than the 58 ‘no gain’ cases. Our study shows that architectural distortion and non-spiculated high-density masses on the mammogram prior to diagnosis are associated with a possible gain in prognosis. Earlier detection of the carcinomas preceded by these signs may well have an impact on breast cancer mortality and thus warrant extra attention in radiological practice.
Radiology review of the UKCCCR breast screening frequency trial: Potential improvements in sensitivity and lead time of radiological signs
2003, Clinical Radiology
AIM: To review all previous mammograms of breast cancer cases diagnosed during a randomized trial comparing 3 yearly to annual mammographic screening with a view to identifying and quantifying cases that might potentially have been diagnosed earlier. METHODS: Mammograms of 602 breast cancer cases (399 screen-detected and 203 interval cases) were reviewed in chronological order and suspicious radiological features noted for each mammogram, up to and including the diagnostic mammogram. RESULTS: Of the 602 cases, 79 (13%) had features at diagnosis that were visible on previous mammograms, suggesting a sensitivity of interpretation of 87%. A similar proportion of screen-detected (14%) and interval cancers (11%) had signs at diagnosis that were visible on previous mammograms. The potential for improvement was particularly noted for asymmetric density (sensitivity = 77%, average time visible before diagnosis 14 months) and parenchymal deformity/stellate lesion (sensitivity = 81%, average time visible before diagnosis 12 months). CONCLUSION: The highest sensitivity was observed for comedo-type microcalcifications (sensitivity = 97%, average time visible before diagnosis 5 months). By improvements in sensitivity to asymmetric density and parenchymal deformity/stellate lesion, 4% of tumours could have their time of diagnosis advanced substantially. Warren R. M. L. etal. (2003) Clinical Radiology58, 128--132.
Reviewing interval cancers: Time well spent?
2002, Clinical Radiology
OBJECTIVES: To categorize interval cancers, and thus identify false-negatives, following prevalent and incident screens in the Welsh breast screening programme. SETTING: Breast Test Wales (BTW) Llandudno, Cardiff and Swansea breast screening units. METHODS: Five hundred and sixty interval breast cancers identified following negative mammographic screening between 1989 and 1997 were reviewed by eight screening radiologists. The blind review was achieved by mixing the screening films of women who subsequently developed an interval cancer with screen negative films of women who did not develop cancer, in a ratio of 4:1. Another radiologist used patients' symptomatic films to record a reference against which the reviewers' reports of the screening films were compared. Interval cancers were categorized as ‘true’, ‘occult’, ‘false-negative’ or ‘unclassified’ interval cancers or interval cancers with minimal signs, based on the National Health Service breast screening programme (NHSBSP) guidelines. RESULTS: Of the classifiable interval films, 32% were false-negatives, 55% were true intervals and 12% occult. The proportion of false-negatives following incident screens was half that following prevalent screens (P = 0.004). Forty percent of the seed films were recalled by the panel. CONCLUSIONS: Low false-negative interval cancer rates following incident screens (18%) versus prevalent screens (36%) suggest that lower cancer detection rates at incident screens may have resulted from fewer cancers than expected being present, rather than from a failure to detect tumours. The panel method for categorizing interval cancers has significant flaws as the results vary markedly with different protocol and is no more accurate than other, quicker and more timely methods. Gower-Thomas, K.et al. (2002).Clinical Radiology57 , 384–388
The sensitivity of the assessment process in screening mammography
2002, Breast
A follow-up of 3143 women from four screening centres, who were assessed after screening mammography, showed that 371 had breast cancer diagnosed at that assessment and 62 developed a breast cancer in the following 6 years. Of these, 32 were judged to be false-negatives at assessment, with 23 of the 32 at the site originally suspected. The results show that the assessment process is not infallible and that the scale of the problem only becomes clear when an exhaustive search is made for missed cancers.
A pilot evaluation of the R2 image checker system and users' response in the detection of interval breast cancers on previous screening films
2001, Clinical Radiology
AIM: Breast screening radiologists are under considerable pressure to maximize cancer detection and minimize the number of interval cancers. This study used screening films, previously reported as normal, taken before the development of an interval cancer. The aim was to determine the sensitivity and specificity of the R2 Image Checker System and the responses of users in detecting abnormalities in these previous films. MATERIALS AND METHODS: The first R2 Image Checker System in the U.K. was installed in the Canterbury Assessment Centre for evaluation purposes. System performance was compared with the standard protocol for assessing interval cancers used in the NHS Breast Screening Programme (NHSBSP) Three readers examined the previous films blind before switching on R2 and compared the prompts generated with the location of the subsequent interval cancer. RESULTS: The previous screening films for 104 cases generated 134 R2 prompts on the subsequent cancer side and 109 on the non-cancer side. Readers classified 29 cancers as false-negative or minimal signs and R2 correctly prompted in 15 of these cases (52%). At least one reader rejected the correct prompt in five cases. CONCLUSION: From this preliminary study it appears that R2 overprompts normal areas and underprompts some cancers in these difficult cases. A full multi-centre evaluation is needed to assess the possible contribution of R2 to the NHSBSP. Garvican, L. and Field, S. (2001).Clinical Radiology56 , 833–837.

View all citing articles on Scopus

View full text

Original articleThe identification of false negatives in a population of interval cancers: a method for audit of screening mammography

Abstract

Clinical Radiology

Radiographic applications of receiver operating characteristic (ROC) curves

Radiology

A method of comparing the areas under receiver operating characteristic curves derived from the same cases

Radiology

NHS Breast Screening Programme — Review

Lung cancer detected during a screening program using 4-month chest radiographs

Radiology

Original article
The identification of false negatives in a population of interval cancers: a method for audit of screening mammography