Clinical outcomes and prognostic factors in patients with breast diffuse large B cell lymphoma; Consortium for Improving Survival of Lymphoma (CISL) study

Approximately one-third of non-Hodgkin lymphoma (NHL) arises primarily from sites other than lymph nodes; for this reason, they are usually named as primary extranodal lymphoma [1]. The definition of primary extranodal lymphoma remains controversial, especially in patients where both nodal and extranodal sites are involved. In 1961, a strict definition for primary extranodal lymphoma was first proposed by Dawson et al. [2] for primary gastrointestinal lymphoma, in which the primary disease was confined to the gastrointestinal tract with involvement of no more than regional lymph nodes. Since then, some authors have suggested that a strict definition of primary extranodal lymphoma, which excluded patients with distant disease from the primary extranodal site, may present an incomplete picture of extranodal lymphoma [3‐5]. Hence, more liberal criteria for primary gastrointestinal lymphoma were proposed by other investigators. Lewin et al. [3] described a series of patients with primary gastrointestinal lymphoma, which included patients with contiguous other organ involvement and distant nodal involvement as well as the gastrointestinal tract, provided that the extranodal disease was just the presenting site and constituted the predominant disease bulk [6].

Breast lymphoma has been defined as lymphoma involving the breast and historically has been classified into primary breast lymphoma (PBL) and secondary breast lymphoma (SBL). The criteria for PBL have been a controversial, similar to those of other extranodal lymphomas. In 1972, Wiseman and Liao first proposed criteria for PBL as follows: (1) adequate pathologic specimens; (2) mammary tissue and lymphomatous infiltrate in close association; (3) no prior diagnosis of an extramammary lymphoma; and (4) no evidence of concurrent widespread disease except ipsilateral axillary lymph nodes [7]. All patients not meeting these four criteria were considered to have SBL. According to these criteria, all patients with PBL had localized disease with or without regional lymph node involvement.

This strict definition for PBL has been adopted in most subsequent clinical studies or case reports for this rare extranodal lymphoma. However, the clinical study that provided data in developing the defining criteria had several limitations. The study included a total of 31 patients and evaluated only 16 patients with PBL. This, we thought, was an extremely small number of patients to define criteria for classification. In addition, differences in clinical parameters and treatment outcomes between PBL and SBL were not described. As mentioned above, some investigators asserted that these strict criteria did not reflect the biologic behavior of the disease, rather only representing a localized manifestation of a different primary extranodal lymphoma [3‐5]. In fact, the traditional strict criteria adopted the pattern of spread that was seen in epithelial breast tumors, from the breast to the regional lymph nodes, even though the pattern of spread in NHL might be different from that of epithelial breast tumors. These findings strongly suggest that the traditional strict criteria for breast lymphoma are not based on scientific considerations about the tumor, but on arbitrary decisions. Nevertheless, until recently, there has been a lack of research investigating more scientific criteria that reflect differences in treatment outcomes and prognosis in patients with NHL and breast involvement.

Therefore, we conducted this retrospective study to investigate scientific classification criteria of primary and secondary breast diffuse large B cell lymphoma (DLBCL) that would reflect treatment outcomes and prognosis. We also evaluated the treatment outcomes of patients with PBL according to the new suggested criteria in terms of progression-free survival (PFS) and OS.

PBL is a rare clinical presentation of NHL with a prevalence of 0.04% to 1.1% of all breast tumors and 1.7% to 2.2% of all extranodal NHL [10]. However, the definition of PBL is still unclear. Criteria for PBL were first proposed by Wiseman and Liao in 1972 [7], which were ambiguous and arbitrary criteria that did not consider the biologic behavior of NHL and were not validated as to prognostic significance.

It was an interesting finding that rituximab was not associated with improved outcome in our cohort. It might be related with the fact that the MED group received rituximab more frequently than OED group. In our country, rituximab was available since 2003. From 1994 to 2002, twenty-one patients were consecutively enrolled to this study, and they, of course, were treated without rituximab. Of these patients, 19 patients were the OED group, whilst only 2 patients were the MED group (p = 0.039). This might explain why the MED group received rituximab more frequently than OED group. On the long-term results of pivotal trial about rituximab conducted by the Groupe d'Etude des Lymphomes de l'Adulte (GELA), immunochemotherapy with rituximab showed the significant superiority of OS over chemotherapy alone in low-risk patients according to age-adjusted IPI [11]. However, the difference of OS in high-risk patients was not evident. The MED group, in our analysis, was absolutely poor-risk group because about two-thirds of patients were high-intermediate to high risk according to IPI. Therefore, the lack of impact on OS of rituximab might be associated with fact that rituximab was more frequently used in poor-risk patients.

The prognostic relevance of the IPI in patients with breast lymphoma has been validated in several studies [12, 13]. The IPI in our population was also an independent prognostic factor for PFS but not for OS. Our study had relatively short follow-up duration (median 41.5 months). As a result, only 10 (20.4%) patients in the PBL group had died by the time of analysis. Because of this point, further follow-up is needed to evaluate the influence of IPI on OS.

Recently, the International Extranodal Lymphoma Study Group (IELSG) reported the results of the largest international survey to date of 204 patients with primary localized DLBCL of the breast, in which the study adopted traditional strict criteria for PBL [12]. When we compared the survival of our PBL group, which indicated OED group according to the number of involved extranodal organ, with outcomes seen in the IELSG survey, 5-year PFS (64.9%) and OS (74.3%) in our cohort were superior to those in IELSG survey (54% and 63%, respectively), even though direct comparison is difficult because of the different definitions of PBL. The relatively good survival in our PBL group may reflect different patient characteristics, compared with those from the IELSG survey. The median age in our cohort (45 years) was younger than that in the IELSG survey (64 years), and the proportion of patients with low to low-intermediate IPI in our cohort (89.8%) was higher than in the IELSG cohort (78%). The patients in our cohort may have more favorable prognostic indices, compared with those in the IELSG survey, and this was associated with more favorable survival even though the patients with more advanced stages were included in our PBL group.

Our study also confirmed the previously reported high rate of extranodal recurrence in the PBL [10, 12‐17]. The sites of progression in the OED group were mainly extranodal. Only one patient had progression in lymph nodes without extranodal disease. One of the notable findings in our analysis was relatively high risk of CNS progression in the OED group, even with relatively short follow-up duration. In our analysis, CNS progression occurred in 10.2% of patients in the PBL group. Although considerably lower CNS progression (5%) occurred in the IELSG survey [12], other PBL series consistently showed more than 10% of CNS progression [10, 13‐15, 17]. Furthermore, other extranodal primary DLBCL, such as primary testicular DLBCL, has a high risk of CNS progression of approximately 15% up to10 years [18]. Concerns about late CNS progression in patients with extranodal lymphoma, in contrast to early relapse within 1 year in patients with aggressive nodal lymphoma, were raised in this primary testicular series [18]. Although it is not clear whether late CNS progression is a unique characteristic of primary testicular lymphoma or a common feature of extranodal lymphoma, we cannot exclude the possibility of further CNS progression in our cohort. Additionally, it is an interesting finding that CNS progression did not occur in the five patients who received prophylactic intrathecal chemotherapy. Although the value of prophylactic intrathecal chemotherapy could not be evaluated in this analysis because of the small number of patients, we thought this may be a meaningful finding when we considered that no CNS progression occurred in a subgroup of patients who received prophylactic intrathecal chemotherapy in the IELSG survey as well [12]. Thus, further studies are needed to provide solid evidence to support the use of prophylactic intrathecal chemotherapy in patients with PBL.

There is still no universal standard treatment of PBL. Most studies agree that the combination of limited surgery, anthracycline-based systemic chemotherapy, and radiotherapy is the best therapeutic option in patients with PBL [12, 13, 17]. Interestingly, more than four cycles of anthracycline-based systemic chemotherapy was associated with improved PFS and OS in the PBL group, whereas radiation therapy was not associated with improved PFS and OS. This observation may be quite different from previous PBL series, which reported that radiation therapy was associated with survival benefit [12, 13]. However, our results about radiation therapy should be interpreted with caution because a high proportion of patients (61.9%) in PBL group were treated with modified radical mastectomy. Several studies reported that survival was not improved by surgery in patients with PBL [12, 16, 19]. Furthermore, the IELSG reported that radical mastectomy, which caused delayed initiation of systemic chemotherapy, was associated with poorer survival [12]. In our study, the proportion of patients treated with modified radical mastectomy was more than twice that of the IELSG survey (30%), and this factor might offset the merits of radiation therapy. Also, an abbreviated course of systemic chemotherapy as well as extensive surgery might influence the poorer survival in our analysis. A randomized trial, comparing a full course of cyclophosphamide, doxorubicin, vincristine, and prednisolone (CHOP) alone with an abbreviated course of CHOP followed by involved-field radiotherapy (IFRT) in patients with nonbulky, stage I-II, nodal DLBCL, was conducted by the Southwest Oncology Group, and the abbreviated course of CHOP plus radiotherapy was superior to full course CHOP alone, with median follow-up of 4.4 years [20]. However, an update of that study with longer follow-up showed convergence of survival curves as a result of an excess of relapses and deaths from lymphoma in the group given the abbreviated course of CHOP plus IFRT [21]. The authors explained that these results might be associated with the inclusion of patients with adverse prognostic factors, such as elevated LDH levels seen in 20% of patients. Thus, the Southwest Oncology Group study emphasized the significance of systemic chemotherapy in patients with adverse prognostic factors. Our analysis included about one-third of patients with elevated levels of LDH in the OED group, and it is likely that an abbreviated course of systemic chemotherapy negatively influenced the outcome in these patients.

Generally, the addition of rituximab to chemotherapy has improved of survival in patients with aggressive nodal DLBCL [22, 23]. However, the impact on survival of rituximab has never been studied in patients with PBL. Persky et al. [24] reported that the addition of rituximab to three cycles of CHOP plus IFRT in patients with high-risk, limited-stage DLBCL, resulted in modest improvement of PFS and OS compared with historical cohort. In that study, high-risk patients were defined as those with at least one adverse risk factor by the stage-modified IPI (nonbulky stage II disease, age >60 years, WHO performance status of 2, or elevated serum LDH). However, this improvement was smaller than observed in advanced disease, and they hypothesized that the modest improvement might be related with a unique biology of limited-stage DLBCL contrast to advanced-stage disease. This finding was meaningful for us to infer the role of rituximab in patients with PBL. More than two-thirds (71.4%) in the OED group of our cohort were high-risk, limited-stage disease. And thus, it was likely that rituximab in patients with PBL might add smaller benefit than as we expected. Furthermore, there was a report suggesting that survival benefits with rituximab were not evident in patients with PBL, in which patients were treated with rituximab and dose-dense chemotherapy (CEOP14) and compared with historical controls [25]. In our analysis of OED group, they did not show a significant PFS difference between patients with chemotherapy plus rituximab and those with chemotherapy alone, but only showed a trend of improved OS for patients treated with chemotherapy plus rituximab. In contrast to nodal, advanced DLBCL, limited role of immunochemotherapy with rituximab would be suggested in this rare extranodal DLBCL. Thus, even though success of rituximab in nodal, advanced-stage DLBCL, further studies are needed for demonstrating the efficacy of rituximab in patients with PBL.

The limitations of the current study include the followings. First, the small number of patients, heterogeneity of the population and short follow-up duration are important limitations of this study. Second, molecular prognostic phenotypes such as germinal center (GC) and activated B cell were not included in this analysis because study protocol approved by IRBs of each center did not require mandatory tissue collection. Recently, some studies have reported that GC phenotype of DLBCL have a significantly better outcome compared to patients with the non-GC phenotype [26, 27]. However, other studies suggested that the addition of rituximab to CHOP-like chemotherapy was associated with overcoming the adverse prognostic impact of non-GC phenotype [28, 29]. About two-thirds of patients (61.8%) in our analysis were treated with rituximab-containing regimens. Therefore, it was likely that the influence of molecular phenotypes on prognosis would quite be decreased in our cohort. Nonetheless, it is major limitation that the results of molecular phenotypes on this series of patients are not available.

As we stated above, several questions about treatment strategies of PBL were raised in this retrospective analysis. Solid evidence, as we know, should come from the results of randomized trials designed on the basis of findings from retrospective studies. The rarity of the disease, however, makes randomized trials virtually impossible in a single institution or nation. Hence, international collaboration is required to adequately conduct randomized trials. Future trials for PBL should consider adopting the criteria according to the number of involved extranodal organ, and include questions raised in our study, including the role of rituximab, prophylactic intrathecal chemotherapy, and radiation therapy. The results of such trials can provide evidence to define the best treatment strategies for PBL.

In conclusion, our analysis showed classification criteria for PBL reflecting differences in treatment outcomes, prognosis and patterns of progression, compared with the traditional strict criteria. A full course of anthracycline-containing systemic chemotherapy appeared to be the most important treatment for patients with PBL. However, we also found that extensive radical surgery might be harmful and therefore should be avoided. Nevertheless, extranodal progression, especially in the breast and CNS, is the main reason for treatment failure in patients with PBL, resulting in shorter long-term survival. Further studies of prospective design with international collaboration are warranted to evaluate the role of rituximab, prophylactic intrathecal chemotherapy, and radiation therapy in patients with PBL according to our suggested criteria.