Invasive breast carcinoma is one of the most frequent types of malignant neoplasm and the leading cause of cancer-associated mortality in females (1). The clinicopathological characteristics typically used to assess prognosis and guide therapy for this disease frequently fail to characterize the multiple aspects and clinical heterogeneity of breast cancer, particularly concerning the biological behaviour of tumours in individual cases.

DNA flow cytometry is a widely-used technique to estimate the DNA content (DNA ploidy) and proliferative activity (S-phase fraction; SPF) in a number of human neoplasms. In breast cancer, DNA ploidy and SPF have been investigated in association with the biological characterization of tumours, patient survival and prognosis (2–15).

Numerous studies have reported that DNA ploidy and/or SPF are prognostic markers for breast cancer (2–11); however, a number of studies have questioned their independent prognostic significance (12–15). These conflicting data have limited the prognostic value of DNA flow cytometry and the extent of its clinical use. The primary reason for the discrepant data appears to be a lack of consistent standardization and quality control programs to ensure the intra- and inter-laboratory reproducibility of results (5,16–20). In addition, intratumoral biological heterogeneity (21,22) and SPF categorization (16,17) have been implicated in the variability of the results. However, despite the inconclusive data on the prognostic value of DNA flow cytometry, numerous hospitals and cancer centres perform cytometric analyses to aid in the measurement of prognostic markers for breast cancer (23).

Previous studies have indicated that the evaluation of cytometric parameters can improve the management of breast cancer, aiding in the prognostic assessment and therapeutic planning for patients (2–6,9,10,24). The current study aimed to review the application of DNA flow cytometry in the prognostic assessment of patients with breast cancer. The present review is based on a 25-year experience of using DNA cytometric technology at the Portuguese Institute of Oncology of Lisbon (Lisbon, Portugal).

A consistent result from our previous cytometric studies was the significant correlation between DNA aneuploidy/high SPF and a high grade of tumour differentiation (P<0.001) (7,8,11), greater tumour size (P<0.002) (7), lack of hormone (oestrogen and progesterone) receptors (P<0.001) (7,8,11) and human epidermal growth factor expression-2 (HER-2; P<0.001) (27). In addition, no significant association between DNA ploidy/SPF and axillary lymph node status was identified (7,8,11,27). Relevant findings are summarized in Table I.

In a study of 308 patients with breast cancer (7), with a short-term follow-up (median follow-up time, 39.6 months), Cox's proportional hazard regression model analysis was used to investigate the disease staging (tumour-node-metastasis) (28), histopathological tumour grading (Nottingham Histologic Score system) (29), DNA flow cytometry results, oestrogen and progesterone receptor expression status, and marker of proliferation Ki-67 (Ki-67) and HER-2 immunohistochemical expression status. Univariate analysis (Cox's proportional hazard regression model analysis) revealed that there was a significant difference between all the above mentioned prognostic factors, excluding Ki-67 expression status (7). Multivariate analysis (Wald test) demonstrated that axillary lymph node status (OS, P=0.001; DFS, P<0.001), DNA ploidy [OS relative risk (RR), 3.43; P=0.006], and SPF (DFS RR, 4.53; P<0.001) retained independent prognostic significance (7).

In a subset analysis performed in a previous study (7), aneuploidy (P=0.005) and a high SPF (P<0.001) were significantly associated with poor DFS and OS survival in patients with axillary node-positive breast cancer. No significant differences were identified between a higher frequency of aneuploidy/a high SPF and a shorter survival time in patients with axillary node-negative breast cancer (7).

In several of our previous studies (7,11,30), patients with aneuploidy tumours, including hyperdiploid/tetraploid tumours (OS RR, 1.96; P=0.003; DFS RR, 1.70; P=0.006) and multiploid/hypertetraploid tumours (OS RR, 2.91; P<0.001; DFS RR, 2.31; P=0.003) demonstrated a significantly poorer prognosis compared with patients with DNA diploid tumours (11). In addition, patients with multiploid/hypertetraploid tumours (Fig. 1) had an increase in tumour biological aggressiveness and a poor clinical evolution compared with patients with diploid and hyperdiploid/tetraploid tumours. To investigate the underlying genetic mechanisms associated with the development of this aneuploidy, copy number variations in a cohort of patients with multiploid/hypertetraploid tumours (n=13) were analysed using comparative genomic hybridization (31). A high incidence of genetic aberrations, primarily associated with chromosomal gains was identified in this group and the 7q locus appeared to be a potential site for a number of candidate oncogenes (31).

In a study of 306 patients with primary operable breast cancer (disease stages I and II), flow cytometry and immunohistochemistry analysis were simultaneously performed in 230 cases (25). Cut-off points of 6.1% for SPF and 10% for Ki-67 expression were used to discriminate between tumours with low and high proliferation. An agreement in the proliferation data between the two methods was identified in 69.1% of cases (K statistic=0.384). However, as a cell proliferation variable SPF, but not Ki-67 expression, demonstrated prognostic significance in regards to OS (P<0.0003) and DFS (P<0.0001) (25). Therefore, SPF appears to be a better prognostic indicator for breast cancer compared with Ki-67 expression (25).

In a prospective study (8) involving 392 patients with advanced breast cancer (stage IIB, n=106; stage IIIA, n=66; stage IIIB, n=174; stage IV, n=46) and a median follow-up time of 81 months, multivariate analysis (Cox's proportional hazard regression model analysis; Wald test) of OS using FNAC revealed that an advanced disease stage (P<0.001), the absence of progesterone receptors (P=0.008) and DNA aneuploidy (P=0.016) had independent prognostic significance (8).

Due to previous results regarding the prognostic value of DNA flow cytometry in patients with breast cancer with a short-term follow-up (7), a study of 393 patients with invasive ductal breast carcinoma with a long-term follow-up (median follow-up time, 134 months) (11) was performed to evaluate ‘classical’ and cytometric prognostic parameters. In multivariate analysis (Cox's proportional hazard regression model analysis; Wald test), axillary node involvement (OS and DFS, P<0.001), DNA aneuploidy (OS, P=0.002; DFS, P=0.001) and lack of progesterone receptor expression (OS, P=0.029) demonstrated statistically significant associations with a shorter patient survival time. In axillary node-negative patients, DNA aneuploidy (DFS, P=0.030) and absence of oestrogen receptor expression (OS, P=0.010) significantly predicted poor survival. In axillary node-positive patients (OS, P=0.010; DFS, P=0.051) and those with moderately differentiated tumours [grade (G)2; OS, P=0.045; DFS, P=0.011] aneuploidy was an indicator of poorer prognosis.

The results of a previous study (11) led to an investigation into the association between the grade of tumour differentiation and DNA ploidy in a large cohort of 684 patients with invasive breast carcinoma with a long-term follow-up (32). Patients with aneuploid G2 tumours demonstrated significantly poorer DFS (P=0.001) and OS (P<0.001), and patients with aneuploid G1 tumours demonstrated significantly poorer OS (P=0.013) (Fig. 2), compared with patients with G2 and G1 diploid tumours, respectively. The same significant prognostic associations were identified in a subset of patients (n=451) with stage I/IIA breast cancer (aneuploid G2 tumour DFS, P=0.003; OS, P=0.004; aneuploid G1 tumour OS, P=0.049). These results suggest that aneuploidy identifies subsets of patients with well (G1) and moderately (G2) differentiated tumours with a poor clinical outcome.

Numerous prognosis-associated studies using DNA flow cytometry have been performed in breast cancer (2–15); however, the results are notably variable. Contradictory findings may result from differences in study design and technical, analytical and interpretative methods (15,17–20). Despite these limitations, the prognostic significance of DNA ploidy and SPF in breast cancer is relatively consensual, at least in univariate models (3,10,33). However, an important question remains in relation to the clinical utility and predictive value of DNA ploidy and SPF in the selection of adjuvant therapies.

Similarly to studies by other institutes (34,35), data from the Portuguese Institute of Oncology of Lisbon indicates that when DNA flow cytometric analysis is performed on fresh/frozen tumour samples from a large cohort, DNA aneuploidy and/or high SPF are significant poor prognostic factors for breast cancer, as determined by univariate and multivariate analyses. Michels et al (35) and Wenger and Clark (36), in their comprehensive reviews of studies involving >100 patients and using fresh or frozen material, demonstrated that 21/24 studies, with 17 using multivariate analysis, reported an association between high SPF and poor patient outcome. Regarding DNA aneuploidy, 17/23 studies, with 8 of them using multivariate analysis, identified a correlation with poor patient prognosis. However, a number of studies using state-of-the-art DNA flow cytometry, in regards to sampling size, recommended protocols and data analysis, have reported that SPF and DNA ploidy do not provide significant prognostic information in breast cancer (13,15).

Data from the Portuguese Institute of Oncology of Lisbon demonstrates that DNA aneuploidy and a high SPF, markers of genetic instability and cell proliferation, respectively, are significantly associated with established molecular features of poor patient prognosis, including a high histological grade, and the absence of hormone receptor and HER-2 expression (20,27,35). Nevertheless, the significant association of DNA aneuploidy and a high SPF with tumour size, but lack of association with lymph node status suggests that they are indicators of tumour growth rather than metastatic potential (15,37).

In our previous studies (7,11), DNA ploidy had a significant prognostic influence in short and long follow-up periods, similar to the results of another study (13). This prognostic influence was most evident in axillary node-positive patients, a subgroup with a poor prognosis. However, in the long-term follow-up study (11), DNA ploidy retained an independent prognostic value for DFS in the subset of patients with axillary node-negative breast cancer, a subset for whom it is difficult to identify significant prognostic indicators for. This finding has been corroborated by others (33,34). Therefore, it is important that DNA ploidy can aid in the prediction of recurrence risk of patients with axillary node-negative breast cancer.

Similarly to the early stages of breast cancer, the prognostic value of DNA ploidy in advanced breast cancer, using FNAC, was identified to be significant (8). Patients with diploid tumours, compared with those with aneuploid tumours, exhibited a more positive OS (8). Notably, despite the screening mammography programs implemented, there are a non-negligible number of cases of locally advanced and metastatic breast cancer, diagnosed ab initio, that still constitute a number of oncological mortality cases. FNAC has the advantage of overcoming a number of sampling disaggregation problems associated with DNA flow cytometry (38).

Notably, the subgroup of patients with relatively uncommon multiploid/hypertetraploid tumours, comprising 5–10% of all cases, had the poorest clinical outcome (7,11,30). Other studies that investigated aneuploid subsets of patients with breast cancer have demonstrated similar results (3,39,40). It is reasonable to hypothesise that the occurrence of multiple aneuploid clones and/or hypertetraploidy supports the development of a more aggressive phenotype (41). Supporting this, Duesberg et al (42) reported that the degree of genomic instability is proportional to the degree of aneuploidy, potentially leading to tumour metastasis.

The association between SPF and DNA ploidy, particularly diploidy with low SPF and aneuploidy with high SPF, is well known. Therefore, the mutual inclusion of DNA ploidy and SPF in multivariate Cox's proportional hazard regression model analysis can produce confounding results. The association between SPF and DNA ploidy can explain why, in our previous short-term follow-up study (7), DNA ploidy and SPF demonstrated independent, prognostic significance for OS and DFS, respectively. A number of studies have recommended that an adequate SPF adjustment should be made in order to eliminate the redundant association with DNA ploidy and improve the prognostic model (43). Despite their similar prognostic significance, there are three primary reasons that make DNA ploidy a more suitable cytometric parameter in routine practice: i) The possibility of assessing ploidy in formalin-fixed paraffin-embedded sample; ii) the technical inability to determine SPF in ~1/4 of cases; and iii) difficulties in the reproducibility of SPF measurements.

Even with inherent technical limitations, the importance of SPF as a prognostic cell proliferation variable cannot be underestimated. Numerous studies have reported that SPF is the most important predictor of survival in breast cancer (2,9,10,16,17). In the studies performed at the Portuguese Institute of Oncology of Lisbon, besides the prognostic significance of SPF in the subgroup of axillary node-positive patients (7), a comparison between SPF and Ki-67 expression for assessing tumour proliferative activity (25) revealed that only SPF had statistical significance for predicting disease outcome. Ki-67 expression analysis has the technical advantage of allowing the morphological evaluation of proliferating cell populations; however, there is still an important unresolved issue concerning the identification of the optimal Ki-67 cut-off value for discriminating between tumours with low and high proliferation (44,45).

Similarly to numerous studies (10,19,33,34,37,46), results from the Portuguese Institute of Oncology of Lisbon have consistently identified a strong association between the grade of tumour differentiation and DNA ploidy in breast cancer (11,32). This result is explained by the similar biological significance of these characteristics, reflecting the extent, type and complexity of genomic abnormalities of the tumour (38,47). In addition, the majority of well-differentiated G1 tumours are diploid and the majority of poorly differentiated G3 tumours are aneuploid (11,32,46). This supports the usefulness of DNA ploidy analysis as a complementary or alternative tool to conventional histological tumour grading, providing additional and less subjective prognostic information. Furthermore, the finding that aneuploidy can identify subgroups of patients with G1 and G2 tumours that have poor prognosis increases the significance of the clinical application of DNA ploidy determination (32). The clinical relevance of this finding is associated with the potential rationally-based influence of DNA ploidy on the therapeutic decisions for patients with aneuploid low/intermediate grade tumours, particularly those whose tumours are luminal A or luminal B/HER-2-negative endocrine-responsive breast cancer, who may benefit from adjuvant chemotherapies.

In conclusion, despite the growing clinical interest and potential implications for therapeutics, the majority of complex high-throughput molecular/genomic technologies (48) are not yet fully standardized or adapted to routine practice (49,50). Notably, relevant prognostic information for breast cancer could easily be obtained by relatively simple and fast DNA flow cytometry, even in cost:benefit financial terms. Investigations into the prognostic significance of DNA flow cytometric analysis for different molecular subtypes of breast cancer, including luminal A, luminal B, HER-2 positive and triple-negative, in the future will prove interesting. Although the studies performed at the Portuguese Institute of Oncology of Lisbon are robust and reliable, they are retrospective in nature. Therefore, further well-designed randomized clinical trials with a larger cohort of patients are required to confirm the accuracy and reliability of the findings discussed in the present review.