BACKGROUND. Extant information reveals inconsistencies concerning the
natural history, pathologic features, and treatment of lobular carcino
ma in situ (LCIS) of the breast. It is uncertain whether these are rel
ated to the methods of study, diagnostic criteria employed, relative p
aucity of cases, or varying lengths of follow-up. METHODS. The cohort
was comprised of 182 women with LCIS who were enrolled in National Sur
gical Adjuvant Breast Project (NSABP) Protocol B-17 but received no tr
eatment other than lumpectomy. Nineteen pathologic features were asses
sed and related to ipsilateral breast tumor recurrence (IBTR) and cont
ralateral breast tumor recurrence (CBTR) at a mean time on study of 5
years. RESULTS. Thirteen IBTR and 4 CBTR, including 1 instance of bila
teral recurrence, were observed. All IBTR occurred in the same quadran
t as the index LCIS. All 4 (2.2%) IBTR that were invasive cancers were
of the lobular type, as was 1 of the 2 (1.1%) CBTR that were invasive
. The other was a mucinous carcinoma. Three (1.6%) IBTR were pure duct
al carcinoma in situ (DCIS) and another was accompanied by LCIS. One i
nstance of CBTR was also comprised of DCIS and LCIS. The remaining fiv
e IBTR and one CBTR were LCIS only. The only pathologic parameter foun
d to be significantly predictive for invasive IBTR and DCIS was type 3
and, to a lesser extent, type 2 LCIS. Some heretofore unrecognized or
little appreciated pathologic features of LCIS are noted. Ancillary h
istochemical findings strongly implicate the derivation of LCIS from d
uctal or secretory cells rather than ''new cells'' or myoepithelial el
ements. All examples tested were found to be c-erb B-2 negative, unive
rsally diploid with normoproliferative DNA content, and estrogen recep
tor and progesterone receptor positive. No other events related to the
breast were encountered. CONCLUSIONS. The number of events observed i
n this large cohort of patients with LCIS is markedly less than that n
oted by others after a comparable period of follow-up. Possible reason
s for this dichotomy, including differences in patient characteristics
, diagnostic criteria, and status of resection margins, are discussed.
Considerations are also offered to support the view that LCIS may exh
ibit precursor activity as well as represent a risk factor (the term m
arker is literally inaccurate). In this light, the designation LCIS ra
ther than lobular neoplasia is preferred. These preliminary findings a
nd historical information presented in this study fail to provide any
reason to perform mastectomy on patients with LCIS. (C) 1996 American
Cancer Society.