|Home | About | Journals | Submit | Contact Us | Français|
Understanding the molecular factors that distinguish inflammatory breast cancer (IBC) from non-IBC is important for IBC diagnosis. We reviewed the records of 48 IBC patients and 64 non-IBC patients from Egypt. We determined RhoC expression and tumor emboli and their relationship to demographic and reproductive characteristics. Compared with non-IBC patients, IBC patients had significantly lower parity (P = 0.018) and fewer palpable tumors (P < 0.0001). IBC tumors showed RhoC overexpression more frequently than non-IBC tumors (87% vs. 17%, respectively) (P < 0.0001). Tumor emboli were significantly more frequent in IBC tumors than non-IBC tumors (Mean ± SD: 14.1 ± 14.0 vs. 7.0 ± 12.9, respectively) (P < 0.0001). This study illustrates that RhoC overexpression and tumor emboli are more frequent in tumors of IBC relative to non-IBC from Egypt. Future studies should focus on relating epidemiologic factors to molecular features of IBC in this population.
Inflammatory breast cancer (IBC) is the most lethal form of breast cancer. In the U.S., the 3-year survival rate of patients with IBC is 42% versus 85% for those with other forms of breast cancer.1 IBC is an exceptionally aggressive disease with rapid onset and distinctive clinical characteristics, including edema, redness, dimpling of the skin, and a low frequency of occurrence of tumor masses.2 There are no histopathologic diagnostic criteria for IBC and the disease is mainly diagnosed based on the clinical presentation-a procedure that is not standardized worldwide.3,4
Research on IBC patients in the U.S. demonstrates that there are certain molecular factors that seem to characterize IBC tumors.5 One such factor is RhoC guanosine triphosphatase (GTPase), a member of the Ras-super family that is involved in cytoskeletal reorganization.6 RhoC is typified as a transforming oncogene for human mammary epithelial cells. Using in situ hybridization, our previous study showed overexpression of RhoC mRNA in 90% of IBC tumors and 38% of non-inflammatory breast cancer (non-IBC) tumors in the U.S.7 In addition, the overexpression of the RhoC GTPase gene, and potentially other genetic mechanisms, help contribute to the motile and invasive IBC phenotype.8
IBC occurs in approximately 2% of all breast cancers in the U.S.,9 in Egypt it occurs in approximately 10% of all breast cancers, and exhibits a much more aggressive clinical course, and occurs with a younger age of onset.10 The very grave signs of IBC found in patients from Egypt might be correlated with molecular factors. We reported that the RhoC expression level and the number of tumor emboli were significantly higher in IBC tumors from Egyptian patients than in IBC tumors from U.S. patients.11 However, it is not clear if RhoC overexpression and high numbers of tumor emboli occur exclusively among IBC tumors or if they are characteristics common to all breast cancers found in Egyptian patients. Therefore, in this study we examined whether the molecular characteristics (namely, RhoC over-expression and higher number of tumor emboli) are present in all Egyptian breast cancers or only in IBC tumors.
IBC subjects included in this study were 48 patients newly diagnosed based on clinical criteria of redness, edema, and peau d’orange during the period of September 2005–December 2006 at the National Cancer Institute of Cairo University (NCI-Cairo) (41 patients) and Tanta Cancer Center in Tanta, Egypt (7 patients). The IBC patients were recruited through our clinical pilot study designed to investigate the frequency of IBC in Egypt (unpublished data). This study was a collaboration between the National Cancer Institute (NCI), Bethesda, and NCI-Cairo. Tissue specimens obtained for this study were from biopsies or mastectomies of IBC patients. Two of the authors (SE and CGK) in Egypt and the U.S independently confirmed the histopathology. Histopathologic types included invasive ductal carcinoma (43 patients, 90%), infiltrative lobular carcinoma (2 patients, 4%), and mixed ductal and lobular carcinoma (3 patients, 6%).
We followed a thorough review of the medical records and pathology reports to ensure getting a representative sample of non-IBC patients whose biopsies contained skin to specifically examine for the presence of dermal lymphatics in this group. The 64 non-IBC patients were all the non-IBC patients who fulfilled the inclusion criteria of tumors 5 cm or larger (T3) stage II or III tumors (with no distant metastasis or M0). These criteria were determined before beginning the study and before reviewing the pathology reports or finding the paraffin blocks, so in the process of selection, there was no other source of bias introduced at all. We decided on the criteria that would allow us to answer the scientific question of whether there are significant differences between IBC and non-IBC with regards to the variables studied and included all of the patents who fulfilled these criteria. This is the major unresolved problem in this field, namely how to truly distinguish by histological or molecular features between true rapidly progressing IBC and indolent and neglected locally advanced cancers. This paper is the first one to make a major contribution to proposing some real and objective differences, besides the time course of the development of symptoms, which is of course a crucial element of the clinical history, but unfortunately, it is seldom adequately documented.
Specifically, our review process included examining all 1893 pathology reports of any breast tissue specimen documented at the Department of Pathology of NCI-Cairo for 2005. Of 1562 pathology reports that revealed cancer, 1236 were non-IBC and 326 were duplicate reports. Reports of outside referral cases comprised 325 patients, and, therefore, we excluded these patients because no paraffin blocks were available. Three hundred seventy-nine reports were for biopsies without sufficient tissue in the paraffin blocks for proper analysis and were excluded as well. Since only invasive ductal carcinoma and infiltrative lobular carcinoma were selected for this study, patients were excluded if they had fibroadenoma or fibroadenosis (27 patients), ductal ectasia (24 patients), Paget’s disease (16 patients), fibrocystic disease (11 patients), or any other type of non-adenocarcinoma (56 patients). Additional inclusion criteria for non-IBC patients included having tumors of 5 cm or larger and no metastasis (TNM staging of II or III). Among the 398 non-IBC patients who fulfilled the above listed criteria, 23 patients were indeed found to have metastasis (TNM stage IV), 63 patients had redness and peau d’orange (T4), 41 patients had a tumor size smaller than or equal to 2 cm (T1), and 207 patients had a tumor size between 2 and 5 cm (T2). As a result of all of the above exclusions due to planned selection criteria, 64 non-IBC patients were identified and selected for this study. Therefore, the 64 non-IBC patients were all non-IBC patients who fulfilled the criteria of 5 cm or larger stage II or III tumors with no metastasis. These criteria were determined prior to beginning the study and before reviewing the pathology reports or finding the paraffin blocks. The 64 patients were not significantly different from the rest of the non-IBC patients who were stage II or III with biopsies or referred slides with respect to demographic, reproductive, and anthropometric characteristics.
This study was approved by the institutional review boards at the participating institutions in Egypt and the University of Michigan.
Information on sociodemographic and anthropometric characteristics, family history of any type of breast cancer, menstrual and reproductive histories, clinical symptoms, and pathological characteristics were collected by abstracting the data from the medical records and the pathology reports of all the patients included in the study. Variables of interest included age at diagnosis, menopausal status, residence, number of children, body mass index (BMI), family history of any breast cancer, palpable mass at diagnosis, tumor grade, estrogen and progesterone receptors (ER and PR) status, and lymph node involvement. We applied the same exact procedure for abstracting information from the medical records of IBC and non-IBC patients. Abstraction was done by our trained abstractors who participated in our previous studies in Egypt. The clinical diagnosis of IBC was independently confirmed by a panel of oncologists in Egypt (SO, HK, and AH) who are treating the patients and, in addition, before a joint meeting of the collaborators from Egypt and the U.S.
For IBC patients, information on erythema, edema, and peau d’orange at diagnosis was collected and confirmed by 2 junior oncologists who were trained for this study followed by a review of the collected information by 2 of the co-authors (SO and HK). The pathological diagnoses of cancer were confirmed by pathologists in Egypt (SE and HGO) and the U.S. (CGK).
Paraffin-embedded tumor tissues were obtained from NCI-Cairo and Tanta Cancer Center. We then stained the slides with hematoxylin and eosin (H&E) and the polyclonal antibody specific to RhoC.12 All the H&E histology slides of IBC and non-IBC patients were reviewed by at least 2 of 3 co-authors (AL, AG, and CGK) who were blinded to each other’s results. The H&E slides were examined to determine the number of tumor emboli in each slide as reported in our recent study.11
Details of the IHC staining and scoring of RhoC GTPase expression were reported in our previous studies.11–13 To ensure reproducibility and eliminate observation bias, all slides were analyzed by 2 co-authors (AL and AG). The slides were analyzed a third time blindly where the identifying tag was covered and, therefore, one could not discriminate between IBC and non-IBC slides. Each slide was RhoC-scored 3 times to ensure reproducibility. In the few instances where a discrepancy existed between scores, the score from the blind scoring trial was used to help eliminate observation bias. We defined the moderate to strong cytoplasmic stain adherence as indicative of overexpression of RhoC (scores 3 and 4) while no staining or faint staining reflected low levels of expression of the RhoC gene (scores 1 and 2).11,13
Variables were statistically analyzed using SAS version 9.1. Differences in frequencies of descriptive variables and molecular factors between the IBC and non-IBC patients were evaluated by Mantel–Haenszel chi-square tests and Fisher’s exact tests. BMI was categorized according to the WHO classification of BMI in adulthood. 14 The Mann–Whitney U-testwas used to evaluate differences in age and number of tumor emboli between IBC and non-IBC patients. To evaluate the correlation between RhoC expression level and number of tumor emboli, we also computed the non-parametric one-way ANOVA tests for both IBC and non-IBC groups. Furthermore, a logistic regression analysis was conducted to determine the predictive factors of IBC, in relation to non-IBC, yielding a crude and adjusted odds ratio (OR) and 95% confidence interval (95% CI). All P-values were two-sided.
The mean [range] age for IBC patients was 46.9 years [28–70], while the mean [range] age was 50.2 years old [32–79] for non-IBC patients (P = 0.108). Compared to non-IBC patients, IBC patients were more likely to live in rural areas (P = 0.073). Parity was significantly lower and the presence of a palpable mass at diagnosis was significantly less frequent in the IBC group than non-IBC (P = 0.018 and P < 0.0001, respectively). Pre-menopausal IBC patients had significantly fewer children than pre-menopausal non-IBC patients (P = 0.044, data not shown), while the number of children was not different between IBC and non-IBC patients who were post-menopausal (P = 0.243, data not shown). Data collected on menopausal status, BMI, family history of breast cancer, tumor grade, ER/PR status, and lymph node involvement are shown in Table 1. None of these factors showed a significant difference between IBC and non-IBC cases.
Of the 46 IBC patients with RhoC results available, 20 (44%), 20 (44%), 5 (10%), and 1 (2%) had a RhoC score of 4, 3, 2, and 1, respectively. Of the 64 non-IBC patients, 4 (6%), 7 (11%), 29 (45%), and 24 (38%) had a RhoC score of 4, 3, 2, and 1, respectively. The level of RhoC expression was significantly higher in IBC tumors than non-IBC tumors (P < 0.0001, Table 2). The analysis of the combined lower expression of RhoC (score 1 and 2) versus the higher expression of RhoC overexpression (score 3 and 4) showed also significantly higher overexpression in IBC compared to non-IBC (P < 0.0001, Table 2).
The analysis of number of tumor emboli showed that 32 (50%), 18 (28%), and 14 (22%) of the non-IBC samples had none or 1 tumor embolus, 2–8 tumor emboli, and 9 or more tumor emboli, respectively, present in their tumors. Among the IBC specimens, 1 (2%), 19 (41%), and 26 (57%) had 1 tumor embolus, 2–8 tumor emboli, and over 9 tumor emboli, respectively, present in their tumors (P < 0.0001, Table 2). The mean number of tumor emboli in the IBC group was 14.1 with a standard deviation (SD) of 14.0, a median of 10.4, and a range of 1–62.2. In the non-IBC group, the mean was 7.0 (SD = 12.9) with a median of 1.5 and a range of 0–57.0.
In the IBC group, we found some evidence suggesting the co-incidence of RhoC overexpression (score 3 and 4) and the presence of all 3 IBC hallmark symptoms (i.e., erythema, edema, and peau d’orange) (P = 0.0088), but the test for trend between RhoC score and number of symptoms yielded non-significant results (P = 0.5109, data not shown). No association was found between the number of tumor emboli and the number of symptoms in the IBC group (P = 0.8204), and the number of tumor emboli was not significantly correlated with the RhoC level in either the IBC or non-IBC group (P = 0.9369 and P = 0.3297, respectively, data not shown).
Logistic regression analysis showed that both RhoC expression and number of tumor emboli were significant predictors of IBC diagnosis, and parity may have a protective effect from IBC. The OR for a RhoC score of 3–4 (relative to 1–2) was 32.1 (95% CI = 11.0–94.2, P < 0.0001, Table 3). As expected, there were more tumor emboli in IBC tumors than in non-IBC tumors. The OR of being diagnosed as IBC relative to non-IBC for a tumor sample with 2–8 emboli was 33.8 (95% CI = 4.2–273.6, P = 0.0010), and the OR for specimen with 9 or more emboli was 59.4 (95% CI = 7.3–482.2, P = 0.0001, Table 3). After adjusting for age, residence, parity, and number of tumor emboli, a higher level of RhoC expression remained as a significant predictor for IBC (OR = 15.4, 95% CI = 3.4–69.4, P = 0.0004, Table 3). Having higher parity was associated with significantly lower IBC risk (after adjusting for age and menopausal status, OR = 0.4, 95% CI = 0.1–0.95 for women who had 3–5 children and OR = 0.2, 95% CI = 0.03–0.9 for women who had 6 or more children relative to women who had 2 or fewer children). However, this association became non-significant after we controlled for RhoC expression level and number of tumor emboli (Table 3).
Following our previous finding of an increase in the RhoC expression level and the number of tumor emboli in IBC patients in Egypt compared to the U.S.,11 this study served to determine if these molecular characteristics differed between IBC and non-IBC cases in Egypt. In this study, we identified 3 interesting observations: (1) tumors of IBC patients had significantly higher RhoC expression than tumors of non-IBC patients, even after adjusting for age, residence, parity, and number of tumor emboli; (2) tumor emboli were more frequent in IBC tumors than non-IBC tumors as expected; and (3) IBC patients tended to have fewer children and were less likely to present with a palpable mass at diagnosis than non-IBC patients. The association between parity and IBC diagnosis was not likely to be related to differences in age or menopausal status of the study population.
Our observation of higher overexpression of RhoC in IBC tumors than in non-IBC tumors in Egyptian patients is in agreement with previous researches.7,13 The combined results from our previous and current studies suggest that IBC tumors from Egyptian patients exhibit higher levels of RhoC overexpression than IBC tumors from U.S. patients or non-IBC tumors from either Egyptian or U.S. patients. Currently, no set of histopathologic criteria are suitable for the diagnosis of IBC and it is routinely diagnosed only by suggestive clinical criteria (erythema, edema, and peau d’orange).3 The ability to detect the level of RhoC expression in patient tissue biopsies may prove to be important tools that improve the diagnosis of IBC. Only the collection of data and specimen with defined guidelines and strict verifiable criteria as explained above would help clarify the diagnosis of IBC.
In addition to molecular factors, we observed a significant decrease in lifetime parity among IBC patients, and this observation was not likely to be explained by age or menopausal status in the present study. Only one previous study that compared data obtained from patients in France and Tunisia examined the relationship between parity and IBC status, but the results were not significant in either group.15 Other reproductive factors, such as prolonged breastfeeding and younger age at menarche were associated with an increased risk of IBC,15,16 but the information from these earlier studies was not available for use in the current study.
Although all of the non-IBC patients had a palpable mass at the time of diagnosis, the percentage of IBC patients having a palpable mass was also high. We found that about 73% of the Egyptian IBC patients in this study had a palpable mass at the time of diagnosis, which was generally higher than the proportion in clinical experience in the U.S. (50%).2 The relatively higher proportion of palpable masses at diagnosis in Egyptian IBC patients may be due to the more aggressive course of IBC found in Egypt.10
Most of the descriptive data included in the present study showed no statistically significant difference between the 2 groups. Both the IBC and non-IBC patients had no significant differences in previous family history of any type of breast cancer, similar to other studies in the U.S., France, and Tunisia.15,16 In addition, we did not find a significant association between BMI and IBC status. The current findings on the association between BMI and specific types of breast cancer (IBC vs. non-IBC) are inconclusive. After considering age and menopausal status upon diagnosis, Chang et al. found that overweight or obese (BMI > 26.65 kg/m2) women in the U.S. had an increased risk of IBC (OR = 2.45; 95% CI = 1.05–5.73) compared to non-IBC.16 Obesity (BMI: 30.0+ kg/m2) in postmenopausal women was also associated with poorer survival in IBC patients.17 However, another study in France found a significant crude OR of IBC in obese women only (BMI: 30+ kg/m2: OR = 4.2, 95% CI = 1.3–13.0), but the association was eliminated after taking the duration of breastfeeding into consideration.15
Unlike breast cancer patients in the U.S. and other countries, Egyptian patients exhibit an unusually high proportion of IBC among all breast cancers, a tendency to early onset, a more aggressive clinical presentation, and a characteristic molecular profile.10,11 These features underscore the possibility that particular IBC risk factors are prevalent in Egypt. These may include environmental contaminants or infectious agents such as human mammary tumor virus, which has been associated with IBC tumors in northern Africa.18 Our previous research on non-IBC patients in Egypt suggests that women without a history of lactation had a significantly higher level of serum dichlorodiphenyldichloroethylene (DDE) than women who breastfed (P = 0.002).19 Younger age of onset and older age at first childbirth were also associated with higher levels of serum DDE in premenopausal women.19 Egyptian women living in urban areas (OR =3.1, 95% CI = 1.1–9.3), with infertility (OR = 9.8, 95% CI = 1.1–89.7), and a higher level of oxidative DNA damage, presented with 7,8-dihydro-8-oxo-2′-deoxyguanine (8-oxo-dG) in circulated DNA (OR = 5.8, 95% CI = 1.9–17.5), showed a significantly higher risk of breast cancer than other Egyptian women.20 Interestingly, although the differences were not statistically significant, we found more rural residents in the IBC group than in the non-IBC group in the current study, suggesting a potentially high likelihood of farming-related exposures among IBC patients in Egypt. Further investigations should be conducted to explore the potential environmental exposures and lifestyle factors that may causally correlate with IBC in different countries.
There are many strengths in this study, one of which is the reproducibility of the data. Both the IBC and non-IBC slides were examined multiple times and by multiple researchers to determine RhoC scores and the number of tumor emboli. In the majority of the samples, all evaluations yielded the same results for each specimen. Another strength of this study was that RhoC scoring was done with the researcher blinded to the IBC status of the slide. This helped to remove any potential observation bias and ensured reproducibility. Additionally, this study included a relatively large number of patients with IBC, which is a rare condition. Study participants were selected from a large pool of non-IBC patients who were then matched with IBC patients on histopathologic type, to eliminate any variables that could have possibly mediated or confounded the comparisons between the 2 groups. This is also the first study to investigate epidemiologic and molecular characteristics of IBC and non-IBC in Egyptian patients, who appear to exhibit a unique form of IBC that differs from IBC as seen in patients from the U.S. or other countries.
The study also has limitations. The lack of reproductive data for the IBC and non-IBC patients is a limitation. These data might include the duration of lactation, the number of pregnancies, the time between the disease onset and the most recent pregnancy or lactation, and whether the patient was pregnant or lactating at diagnosis. Data on these reproductive factors were gleaned from the medical records and clinical reports, wherever possible, but this information was often not recorded. We aimed at collecting samples with epidermis for all non-IBC patients by including mastectomies of non-IBC tumors. However, a few of the specimens did not contain full thickness of skin, including Epidermis. Although the sample size of this study was small, we used appropriate statistical tests for such sample size such as the Fisher’s exact test, Mann–Whitney U-test, and non-parametrical ANOVA to robustly evaluate the associations. Although the sample is relatively small, it turns out that the differences are sufficiently robust to reach significant levels; a small sample would not give rise to false positive findings, especially when we are doing just a few comparisons.
The results of this study will give rise to a number of promising future studies. Possibilities include comparisons of molecular components of IBC in different racial and ethnic groups. Additional genetic, lifestyle, and environmental factors should be carefully examined to understand the relationship of RhoC expression and the number of tumor emboli with the diverse risk factors for IBC. Given that the diagnostic criteria for IBC have not been standardized worldwide, international research will shed more light on IBC etiology, and aid in developing a more accurate definition of IBC to be used in diagnosis of the disease.
Grant CA77612, the Burroughs Wellcome Fund, the Breast Cancer Research Foundation, and the Department of Defense, R25 CA112383, K07 CA90241, and R03 CA117350, the University of Michigan’s Cancer Center Support Grant (5 P30 CA46592) from the National Cancer Institute.
Conflict of interest