A basis for updating our approach to resistant acute leukemia†
Conflict of interest: Nothing to report.
Abstract
No studies exist documenting that chemotherapy alone eradicates tumors composed of leukemic cells in a large group of patients with tumors at any one site. Yet, its use has continued over 40 years in the absence of data. Consensus protocols exist only for testis and meningeal tumors, relying on local therapy. To constitute a body of knowledge about tumors at one site, the breast was chosen and all published cases were analyzed, with follow-up obtained, to document the behavior of acute leukemia tumors and survival after presentation. Among 235 cases (52% published since 2000), overall survival was poor, particularly for the 43% with concurrent morphologic marrow relapse, with 66-73% one-year mortality. Only 4 of 106 patients treated with chemotherapy alone survived 4 years. The majority of AML and ALL tumors were only transiently responsive to anti-leukemia treatments, including transplant, and next relapses were as, or more, common in further tumors than in marrow. A pattern of tumors similar to the metastases of invasive lobular breast cancer was revealed. When relapse occurred in marrow, durable remission was only rarely obtained. These data suggest a potential benefit of incorporating extent of disease workup at diagnosis and relapse into prospective trials. This could yield an accurate incidence of extramedullary tumors and a means to identify occult residual disease which could lead to marrow relapse. This approach could potentially result in greater success in curing acute leukemias. Am. J. Hematol. 2012. © 2011 Wiley Periodicals, Inc.
Introduction
Resistant leukemia remains the main cause of treatment failure and mortality for acute leukemia patients. Despite high initial remission rates, leukemia relapses in the majority of adults with AML and ALL [1, 2], at least 20% of children with ALL, and up to 40% with AML [3, 4]. Most relapsed patients die of the disease. Among the means whereby leukemic cells resist chemotherapy is their formation into extramedullary tumors. These may occur in any organ and may be an occult source of systemic relapse more often than is clinically recognized, as body scans and autopsies are not routine. Leukemic tumors in testis and meninges are the only ones for which there are consensus approaches, centered on local tumor-directed therapy, but for extramedullary tumors at all other sites there has been no knowledge base on which to make therapeutic decisions. As a result, treatment with chemotherapy alone has continued over 40 years with no studies confirming its value in eradicating tumors at any one site. Knowledge is limited to individual case reports, but there are now enough reports of leukemic tumors to enable analysis of behavior in each organ. A study was designed to track the course of leukemic breast tumors, enlarging on one which identified 153 cases in 2006 [5], combining those with 50% more cases published in the intervening years, with follow-up obtained from authors. The goal of this study was to track the clinical behavior of these tumors and their patterns of resistance. Recognition of these patterns should lead to new approaches, as marrow-directed leukemia treatment alone has been generally ineffective against leukemic tumors and the mortality for patients who develop these tumors remains very high.
Methods
All cases of leukemic breast tumors were identified in published articles found in Medline®, Scopus®, bibliographies, and meeting abstracts, and authors were contacted for full survival duration. More than two-thirds of authors, from 34 countries, responded and an additional 16 unpublished cases were contributed. Cases not referenced in the text are listed in Supporting Information. The cases were published between 1969 and 2010, 52% since 2000. Data collected from the reports were age, gender, temporal relation of breast tumor to marrow leukemia, other extramedullary site involvement, marrow remission status, types of therapeutic approach, consequences of treatment, and total survival. Obtaining exact treatment doses, details, and timing was neither possible nor the goal of this retrospective study.
“Clinical response” in breast tumor, as determined by the treating physician, was used to denote clinical resolution, instead of “complete remission”, because radiologic confirmation was rare.
For analysis, the cases are divided into the four time periods in which breast tumors present: (1) prior to diagnosis of marrow leukemia, (2) simultaneously with marrow leukemia diagnosis, (3) subsequent to therapy for marrow leukemia, and (4) subsequent to stem cell transplant (autologous or allogeneic). In the latter two groups, the marrow may or may not have been in relapse.
Results
Patient characteristics
Two hundred thirty-five cases were identified. There were 163 cases of AML and 72 of ALL; 13 patients were men. The age of AML patients ranged from 1 to 75 years; 24% were younger than 22. Among ALL patients, ages were 8 to 69; 43% were under 22. All FAB subtypes were represented, none was predominant. Karyotyping was available for only 55 cases of AML (36%) and 18 of ALL (26%). In AML, 16 (29%) were normal, 10 had inv (16), 9 t(8;21), 3 t(15;17), and 6 had +8; 11 other translocations, additions, or deletions were reported in one or two cases each. In ALL, seven (39%) were normal, four had t(9;22), 1 had t(4;11), and six others had various other abnormalities.
Presentation
Most leukemic breast tumors were reported to grow rapidly over days, but many grew over months. Most were symptomatic, but some were found incidentally on self- or physician exam or routine mammograms or scans performed after tumors were found at other sites. The majority of breast tumors were clinically unilateral, as single or multiple nodules or diffuse enlargement; 40% of both AML and ALL cases were clinically bilateral on presentation. These tumors may appear indistinguishable from epithelial breast cancers on mammography and at surgery. Tumor masses varied in size up to 12 cm. In one case, fungating tumor eroded through breast in four sites (Ruiz-Argüelles GJ. Unpublished case. 2010). Axillary lymphadenopathy was found in 64% of 28 AML and 79% of 28 ALL cases where axillary examination was recorded.
For cases of breast tumors presenting subsequent to treatment for marrow leukemia, there was considerable range in the time of onset after diagnosis. In both AML and ALL cases, the median time was 12 months after diagnosis, but some relapses occurred late. Nine AML breast relapses occurred between 3 and 10 years [6-14] and four cases in ALL occurred 3–7 years after diagnosis [15, 16, Milpied N. Unpublished case. 2010, Dutra AP. Unpublished case. 2009]. After transplant for AML (all but four of which were allogeneic), the median time of breast relapse was 16 months, and seven cases were noted after 3–8 years [17-23], including two after autologous transplants. One patient had a leukemic breast tumor 6 years after an allogeneic transplant for severe aplastic anemia [24]. Among ALL patients with breast tumors after transplants (all but two allogeneic), all occurred within 2 years, at a median of 1 year.
Involvement of other extramedullary sites had been clinically evident prior to occurrence of the breast tumor in 12% of AML and 7% of ALL cases. Simultaneous with presentation of the breast tumor, 29% of both AML and ALL cases had clinical involvement of other extramedullary sites, almost exclusively in soft tissues.
Diagnostic studies
The diagnosis of leukemia was made on excisional biopsy or tumor removal in a minority of cases. In the majority, diagnosis was made on core biopsies or needle aspirates. Myeloperoxidase, lysozyme, and PAS staining were commonly used to differentiate leukemic tumors from epithelial breast cancer. Leukemia cell surface markers were recorded in some cases and combined results are shown in Table I.
| AML | ALL | ||
|---|---|---|---|
| CD33 | 11 of 11 | ||
| CD34 | 17 of 21 | CD34 | 6 of 7 |
| CD43 | 16 of 16 | CD43 | 2 of 3 |
| HLA-DR | 7 of 7 | HLA-DR | 5 of 5 |
| CD13 | 8 of 9 | CD3 | 4 of 4 |
| CD43 | 16 of 16 | CD10 | 4 of 5 |
| CD45 | 14 of 14 | CD19 | 4 of 5 |
| CD56 | 5 of 9 | CD20 | 4 of 6 |
| CD68 | 13 of 16 | ||
| CD117 | 8 of 11 | ||
| TdT | 4 of 7 | TdT | 4 of 4 |
Survival
Table II illustrates the overall patient survival at each year after diagnosis of the breast tumors. Survival of patients without simultaneous marrow leukemia is superior to those who had concurrent marrow relapse in every presentation time. Apart from those patients with breast tumors before marrow leukemia, only 36% of the total AML patients survived 1 year, and only 14% survived 5 years. ALL patients had better one-year survival but, like AML, only 15% were alive at 5 years. More than half of those who survived 5 years had relapsed after their initial postbreast tumor responses.
| Cases | Evaluable | Alive 1 yr | Alive 2 yr | Alive 3 yr | Alive 4 yr | Alive 5 yr | |
|---|---|---|---|---|---|---|---|
| AML survival | |||||||
| Prior to BM | 39 | 35 | 28 (80%) | 22 (63%) | 17 (49%) | 14 (37%) | 11 (31%) |
| At dx | 37 | 32 | 11 (34%) | 4 (13%) | 4 (13%) | 3 (9%) | 2 (6%) |
| After chemo | 45 | 40 | |||||
| -BMr ≤1 m | 19 | 4 (24%) | 1 (5%) | 0 | 0 | 0 | |
| -No BMr | 21 | 9 (43%) | 6 (29%) | 5 (24%) | 3 (14%) | 2 (10%) | |
| After SCT | 43 | 41 | |||||
| -BMr ≤1 m | 9 | 1 (11%) | 0 | 0 | 0 | 0 | |
| -No BMr | 32 | 17 (53%) | 13 (41%) | 9 (28%) | 8 (25%) | 5 (16%) | |
| ALL survival | |||||||
| Prior to BM | 1 | 1 | 1 | 1 | 1 | 1 | 0 |
| At dx | 21 | 16 | 6 (38%) | 3 (19%) | 3 (19%) | 2 (13%) | 2 (13%) |
| After chemo | 26 | 20 | |||||
| -BMr ≤1 m | 8 | 1 (13%) | 0 | 0 | 0 | 0 | |
| -No BMr | 12 | 9 (75%) | 5 (42%) | 3 (25%) | 3 (25%) | 3 (25%) | |
| After SCT | 24 | 22 | |||||
| -BMr ≤1 m | 5 | 3 (60%) | 1 (20%) | 1 (20%) | 0 | 0 | |
| -No BMr | 17 | 12 (71%) | 8 (47%) | 6 (35%) | 6 (35%) | 4 (24%) | |
- Evaluable, survival data available; SCT, stem cell transplant; BMr, bone marrow relapse.
Response to treatment
Acute myeloid leukemia
Table III details the types of treatment given to AML cases of breast tumor without morphologic marrow relapse for which there was evaluable relapse data. Among the 80 cases without simultaneous marrow relapse, local therapy alone was employed most commonly to treat breast tumors found prior to marrow leukemia (known as ‘granulocytic sarcoma’). In 71% of these cases, tumor was removed by mastectomy or excision. With no further therapy, the majority relapsed within 2 years, in marrow with or without an extramedullary site. However, it is noteworthy that three cases remained disease-free 3–6 years before relapse [25-27], having had no chemotherapy, and another, who refused chemotherapy after excision and radiation, was still disease-free at 8 years [28]. Three of nine patients with marrow relapses subsequently received salvage chemotherapy for numerous cycles, and obtained lengthy remissions ongoing at 5, 18, and 26 years [29-31]. Local therapy alone produced no long responses among patients presenting after chemotherapy or transplant and further extramedullary relapse was more common than marrow relapse. There is one extraordinary response to total hysterectomy and oophorectomies in an 18-year-old woman, whose large bilateral ovarian tumors were noted 2 months after bilateral breast tumors. With no chemotherapy, there was a surprising disappearance of the breast tumors, confirmed by mammography at 18 months. Her disease-free survival was reported by her physician to be “many years”, because records are no longer available [32].
| 80 Cases Without Marrow Relapse | 57 Cases With Marrow Relapse | |||||
|---|---|---|---|---|---|---|
| Prior to BM | After chemo | After SCT | At Dx | After chemo | After SCT | |
| Total cases | 39 | 21 | 29 | 37 | 19 | 9 |
| Evaluable | 35 | 19 | 26 | 29 | 19 | 9 |
| Local Rx only | 17 | 3 | 5 | 0 | 0 | 0 |
| Clinical response | 17 | 3 | 5 | |||
| EMr (mo) | 2 (12, 72) | 2 (2,5) | 4 (12,14,23,56) | |||
| BMr (mo) | 9 (2,2,2,2,3,20,23,36,42) | 1 (2) | 1 (4) | |||
| EMr/BMr | 4 (6,6,8,18) | |||||
| DF (yrs) | 2 (1, 8 YRS) | |||||
| Local + chemo | 15 | 6 | 13 | 2 | 5 | 2 |
| Clinical response | 15 | 6 | 12 | 2 | 3 | 2 |
| Lost in CR | 2 | 1 | 1 (5) | |||
| Dead in CR (mo) | 1 (6) | 1 (16) | 1 (6) | |||
| EMr (mo) | 3 (1,16,45) | 5 (4,12,12,12,51) | 1 (16) | |||
| BMr (mo) | 4 (3,5,19,96) | 1 (13) | 1 (18) | 1 (1.5) | 1 (22) | |
| EMr/BMr | 1 (4) | 3 (8,14,20) | ||||
| DF (yrs) | 5 (2,2,4,22,26 yrs) | 3 (<1, 1.4, 4.5 yrs) | 2 (6.5, 12 yrs) | 1 (9 yr) | ||
| Chemo alone | 3 | 9 | 8 | 27 | 14 | 7 |
| Partial response | 6 | 1 | 0 | |||
| Lost | 5 | |||||
| Br NR/BM CR | 1 BMr (8) | 1 EMr (16) | ||||
| Clinical response | 3 | 7 | 1 | 8 | 2 | 0 |
| EMr (mo) | 3 (2,4,10) | 1 (25) | 2 (6,16) | 2 (6,6) | ||
| BMr (mo) | 2 (2,6) | 3 (all <9) | 3 (2,6,8) | |||
| EMr/BMr (mo) | 2 (6,8) | |||||
| DF (yrs) | 1 (10 yrs) | 1 (0.7 yrs) | 1 (4 yrs) | |||
- Br, breast; BMr, bone marrow relapse; DF, disease-free; EMr, extramedullary relapse.
Chemotherapy alone was given to a total of 20 cases in the three groups in Table III, with only one response lasting over 25 months. This was a young woman whose breast tumor expressed high levels of WT-1 mRNA, found also in the otherwise normal-appearing marrow. She received 6 months of intensive chemotherapy with resolution of breast tumors, confirmed by PET scan, with disease-free survival ongoing at 10 years [33].
Combination local and systemic therapy was given to a total of 34 patients in the three groups, of whom 11 obtained disease control for periods from 2 to 26 years [25, 34-43]. Seven of these responses were in patients with granulocytic sarcomas, where there was a relapse in marrow as late as 8 years [38]. After combined therapy, among the 18 cases that relapsed, the first site of relapse was as often in extramedullary sites as in marrow, and half the patients relapsed in ipsi- and/or contralateral breasts.
Table III shows the poor response of the 57 evaluable cases whose breast tumors presented with simultaneous marrow relapse. The majority did not respond to treatment. Most cases were refractory to chemotherapy alone, given to 48; complete responses were seen in only 10 patients, and all but one relapsed within 16 months. One was still disease-free over 4 years [44]. The asynchrony of response to chemotherapy between bone marrow and breast tumors is illustrated by two cases in particular. One presented with bilateral breast tumors and 7% marrow blasts. Intensive chemotherapy produced complete marrow remission after one course, with transient clinical disappearance of tumors in both breasts, but the left breast tumor recurred after one consolidation course, and grew from 5 cm × 4 cm to 10 cm prior to the next consolidation. Resolution of the right breast tumor was confirmed by mammogram and ultrasound, while the left one grew, despite continued chemotherapy. Full bone marrow relapse occurred after 6 months [45]. In another case, bilateral breast tumors and marrow relapse occurred 5 months after diagnosis of AML, after a 2-month marrow remission. Chemotherapy produced marrow remission and clinical resolution of the right breast tumors, which did not recur, but the tumor in the left breast remained and continued to grow. Marrow relapse recurred and remission was again achieved, while the left breast tumor grew. It was ultimately removed 16 months after presentation, at which time abdominal tumors were obvious. Shortly thereafter, marrow relapse and death occurred [46].
Stem cell transplants were part of treatment for 34 AML patients with breast tumors, of which six were autologous. All but four had periods of apparent response. Nineteen relapsed: five first in marrow, 12 in extramedullary sites (eight in breast), and two in both. Five died of transplant-related causes, one was lost to follow-up, and six continued disease-free from 1 to 9 years [39, 43, 47-49, Jakubowski A. Unpublished case. 2005], including one well 2 years after autologous graft [43]. Donor lymphocyte infusions were part of treatment for nine cases whose breast tumors occurred after transplants; there were no lengthy responses.
In summary, follow-up data on the 80 evaluable AML patients with breast tumors without concurrent marrow relapse showed only a few prolonged responses, reflecting the high percentage of further relapses in extramedullary sites as often as in marrow, typically within one year of the breast tumor, despite chemotherapy. A principal site of next relapse was the ipsi- or contralateral breast, in at least 21 cases. For the 57 evaluable cases with concurrent breast and marrow leukemia, the majority of cases were resistant to induction chemotherapy, with less than one-third obtaining responses, and all but two relapsing within 22 months.
Acute lymphoblastic leukemia
The response to treatment of 37 ALL patients with breast tumors without simultaneous marrow relapse is shown in Table IV. For 32, there was evaluable information about response. Only one case presented before marrow involvement. After chemotherapy alone produced clinical breast response, there was relapse in femur in 18 months [25]. In 12 evaluable cases of breast tumors occurring subsequent to chemotherapy, all but one had clinical responses. Two were treated with irradiation only; one was reported to be disease-free at 30 years [50] and the other relapsed in marrow at 13 months [51]. Of five treated with combination therapy, one had no breast response and relapsed in CSF in 12 months [52]. Three had short responses ending in extramedullary relapses [53-55] and one patient continued disease-free at 10 years, after busulfan/cyclophosphamide and autologous transplant followed by breast irradiation [15]. All five treated with chemotherapy alone had clinical responses; three relapsed in extramedullary sites within 8 months [51, 54, 56], one relapsed in breast and marrow at 24 months [57], and one was disease-free at 16+ months [16]. Similar, almost universal, response to therapy was seen among the 19 post-transplant cases (17 allogeneic, two autologous), which were short-lived in most cases. Of the six treated with local therapy alone, one had no obtainable follow-up after radiation [58], one relapsed in marrow while undergoing radiation [59], and four had early extramedullary relapses [60-62, Selleri C. Unpublished case. 2002]. Of the eight patients treated with combination therapy, seven had extramedullary relapses [63-67] and one had no disease recurrence at death from an unrelated cause after 51 months [Nichols G. Unpublished case. 2002]. One of the four responses to chemotherapy alone lasted 4 years [68, 69, Mensah-Glanowska P. Unpublished case. 2005]. In summary, in ALL patients whose breast relapses occurred without marrow relapse, initial response was common, at least 75% relapsed, and extramedullary tissue was the most common site of next relapse, in 21 cases, whereas marrow was the next site in only four.
| 32 Cases without Marrow Relapse | 32 Cases with Marrow Relapse | |||||
|---|---|---|---|---|---|---|
| Prior to BM | After chemo | After SCT | At dx | After chemo | After SCT | |
| Total cases | 1 | 17 | 19 | 21 | 9 | 4 |
| Evaluable | 1 | 12 | 19 | 19 | 9 | 4 |
| Local Rx only | 1 | 2 | 6 | 0 | 0 | 0 |
| Clinical response | 1 | 2 | 6 | |||
| Lost | 1 | |||||
| EMr (mo) | 1 (18) | 4 (1,1,5,6) | ||||
| BMr (mo) | 1 (13) | 1 (1) | ||||
| DF (yrs) | 1 (30+ yrs) | |||||
| Local Rx + Chemo | 0 | 5 | 8 | 2 | 0 | 2 |
| Clinical Response | 4 | 8 | 2 | 2 | ||
| Lost | 1 | |||||
| Died in CR | 1 (51) | 2 (13.7, 37) | ||||
| EMr (mo) | 3 < 6m | 7 (2,5,5,5,31,48,50) | ||||
| BMr (mo) | 1 (1) | |||||
| DF (yrs) | 1 (10+ yrs) | |||||
| Chemo alone | 0 | 5 | 5 | 17 | 9 | 2 |
| Clinical Response | 5 | 4 | 16 | 5 | 0 | |
| Lost | 1 | 5 | ||||
| Died | 3 | 1 | ||||
| EMr (mo) | 3 (2, 2, 8 ) | 2 (10,48) | 4 (2,2,12,23 ) | 1 (13) | ||
| EMr/BMr | 1 (24 ) | 1 (4) | ||||
| BMr (mo) | 1 (2) | 1 (7) | 1 (3) | |||
| DF (yrs) | 1 (1.5 yrs) | 3 (1, 7, 12+ yrs) | 1 (9+ mo) | |||
Treatment of breast tumors in ALL patients that occurred with morphologic marrow relapse had very poor outcomes, documented among 32 evaluable cases. As seen in Table IV, the rare durable responses were seen among the 17 evaluable patients whose breast tumors were noted at the time of ALL diagnosis, treated with chemotherapy alone. Marrow remission and clinical breast response was attained in all but one. Although there were four relapses in extramedullary sites and one in bone marrow, there are three cases disease-free for 1 to 12+ years [70-72]. Chemotherapy alone was not as successful among patients whose breast tumors occurred after a remission induced by chemotherapy; only five of the nine obtained responses, none lasting over 13 months. Among eight patients with breast and marrow relapse after transplant, the only responses were after combination chemotherapy and radiation in two of the post-allogeneic patients, both died in remission with GVHD at 13.7 and 37 months [73, 74].
Other treatments for breast relapse included stem cell transplant, given to 12 patients (10 allogeneic, 2 autologous) and donor lymphocyte infusions given to nine for post-transplant breast relapses. Of those transplanted, one had no response to autologous graft [50] and another, whose t(9;22) leukemia and breast tumors responded to pre-transplant imatinib, died of graft failure [75]. Six relapsed in extramedullary sites, one also in marrow [52, 61, 63, 76, Selleri C. Unpublished case. 2002, Mensah-Glanowska P. Unpublished case. 2005] and four were disease-free at 8 and 16 months and 10 and 12 years after breast relapse [15, 16, 70, Milpied N. Unpublished case. 2010]. Four of the nine patients who received DLI as part of combination treatment in the post-transplant group obtained responses of 13 to 51 months [73, 74, Nichols G. Unpublished case. 2002, Mensah-Glanowska P. Unpublished case. 2002], three of whom died with severe GVHD without relapse.
Relapse sites
The first site of relapse after leukemic breast tumor was available for 84 AML and 34 ALL cases. In AML, it was bone marrow in 41 (including eight with concurrent extramedullary relapses), ipsi- or contralateral breast (including two with marrow) in 21, and 20 relapsed in various non-breast sites (including one also in marrow). In six patients, the central nervous system was the first relapse site and four others had subsequent CNS relapses. Among ALL cases, the first site of relapse after the breast tumor was CNS in 11 patients (three of whom had concurrent skin, breast, or marrow relapses) and five others had CNS relapses subsequently. The ipsi- or contralateral breast was the next site in nine (two were concurrent with CNS). In only eight cases was the next site the marrow (two each with concurrent breast or CNS relapses). Five cases had other non-marrow first relapse sites after the breast: two in sinuses [64, Mensah-Glanowska P. Unpublished case. 2005], and one each in heart [62], femur [25], and retroperitoneum [69]. It is noteworthy that two of the 13 male patients had testicular involvement: an ALL patient with late testicular relapse after several breast recurrences [55], and one AML patient with bilateral testis and breast tumors at diagnosis [77].
Without routine scans, and with autopsies reported for fewer than 5% of the deceased patients, the clinically apparent sites enumerated in the cases in this report are certainly an underrepresentation of organ involvement. Nonetheless, the data suggest that leukemic breast tumors are associated with clinical involvement of CNS (mainly meninges), skin, pelvic, gastrointestinal, and retroperitoneal sites, in both AML and ALL.
Discussion
Important points about the biology of acute leukemia when it forms tumors become apparent from this study of 235 cases, the largest analysis of leukemic tumors at an individual site ever published. They provide a basis for a new perspective on resistant residual disease that could improve prognosis for acute leukemia patients. They include the critical value of local therapy, the elucidation of an independent extramedullary path of tumor progression, the inadequacy of chemotherapy alone, and the similar behavior of AML and ALL.
Local therapy appears to have an important role in eradicating leukemic breast tumors, as it does in breast carcinoma. However, it was used in less than one-third of published cases, principally for de novo breast tumors before morphologic marrow leukemia. Among these are several cases of 3 to 8 year responses without any chemotherapy. This long period of leukemia quiescence raises the question of whether “local myelopoiesis” could occur in some tissues, hypothesized 7 decades ago [78]. These leukemic tumors have typically not been approached by clinicians as solid cancers, but as collections of sensitive cells, which they do not appear to be. Focus has been on the risk of marrow relapse when a leukemic tumor is diagnosed and clinicians may hasten to employ systemic chemotherapy alone, unaware that such tumors are resistant to marrow-directed therapy, that response will likely be transient, and that there is a significant chance of ipsi- and contralateral relapse, followed by further tumor spread. Even when local radiation is given, it has typically been directed only to a single breast. This is in contrast to the approach to unilateral leukemic testis tumors, where bilateral radiation, acknowledging the high risk of contralateral tumors, is standard and effective. The omission of local therapy has been even more common in cases of breast tumors where the marrow is in relapse on presentation, when marrow-directed treatment is urgent. Post-chemotherapy irradiation may be an attractive option in such cases. It was given to five AML and two ALL patients, with responses lasting 2 to 12+ years thereafter [15, 22, 35, 38, 40, 43, 73]; two of these were given radiation after autologous transplants [15, 43]. This study documents that after one extramedullary tumor, there is a high chance that the next site of leukemia will be another tumor in another site, not the bone marrow.
Chemotherapy alone, a commonly reported approach to leukemic breast tumors, has produced few long-term responses in either AML or ALL. There appears to be transient clinical response in most cases, except among the 15 AML cases with breast relapses presenting after stem cell transplants where only one responded. There are only four cases among 106 AML and ALL patients treated with chemotherapy alone with disease-free survivals from 4 to 12 years. They include one AML and two ALL cases who received chemotherapy for newly diagnosed marrow and breast leukemias [44, 70, 72], and the de novo case treated for WT-1 expression in both sites[33]. With these observations, there is little data to justify relying on chemotherapy alone when extramedullary breast tumors, or probably any soft tissue leukemic tumors, are found. The majority of the 46 cases given stem cell transplants after chemotherapy relapsed, but there are seven reports of 2 to 9 year disease-free survivals in both AML and ALL patients after allogeneic or autologous transplants [15, 39, 43, 47, 49, 70 Jakubowski A. Unpublished case. 2005]. Transplant thus offers one of the best current options, provided there are no occult tumors in the patient before initiating conditioning treatment, and that their absence is confirmed after transplant.
Though chemotherapy protocols for AML and ALL are composed of different agents, the results observed suggest their leukemic tumors have similar behavior. After the breast, relapse in another extramedullary site was common, four times more common than marrow relapse among the ALL patients, and equally common in the AML cases. It is noteworthy that in both leukemias, meningeal leukemia developed, in at least 22% of ALL and 6% of AML cases. Although individual prophylaxis details were unobtainable, most ALL protocols include CSF prophylaxis. This supports the value of prolonged CSF monitoring in cases of leukemia in non-marrow sites.
These data suggest that leukemic tumors in extramedullary sites behave more like invasive solid tumors than acute leukemia. The asynchrony of response when anti-leukemia agents are given has been amply documented in this large study. The cases with lengthy responses to local therapy, without chemotherapy, imply that in at least some, the tumor burden may be initially confined to a non-marrow site. The clinical behavior of breast tumors suggests they have the ability to metastasize, independent of marrow relapse, with a proclivity for spread to the contralateral breast, in a pattern paralleling the natural history of invasive lobular breast cancer, which is likewise known to spread to abdomen/pelvis and eventually to meninges [79]. A morphologic similarity of leukemic breast tumors to invasive lobular carcinoma has been noted by some authors [27, 30, 80-82]. The breast, like many organs, is a site of normal hematopoiesis in the embryo through a few weeks after birth [83, 84] and of extramedullary hematopoiesis in myelofibrosis [85], and is occasionally seen in normal infants [86] and cancerous or contralateral breasts after chemotherapy and surgery [87]. Therefore, one could postulate that after an unknown leukemogenic stimulus, the breast (or any other site) could revert to the embryonic state where hematopoietic cells, in this instance malignant ones, could proliferate. Factors in the breast microenvironment may interact with leukemic cells and render them resistant to anti-leukemia agents. It may be that cells trafficking between breast and marrow contribute to the marrow's acquiring resistance to agents to which it should be sensitive, as is illustrated in many of the cases detailed here.
Though fewer than 39% of cases had available FAB or karyotype data, it appears that leukemic breast tumors may occur in patients of any age, FAB, and karyotype. If resistant leukemic tumors cannot be predicted and the prognosis in available cases is so poor, attempting to find tumors early may help us improve results. The potential benefit of extent-of-disease workup for new and relapsed leukemia patients, as is done in every other cancer, could be considered. Incorporating this into prospective trials could help determine the true incidence of extramedullary tumors and their role in the residual disease found in some patients after protocols which eradicate leukemia in others. A large autopsy study suggested that at least 50% of myeloblastic tumors were asymptomatic during life [88], providing reason to suspect that many relapsed patients harbor occult tumors which could be responsible for recurrent marrow disease. As PET, CT, and gallium scans have been useful in finding occult tumors [89-92], we now have the opportunity to alter the slow pace of progress in acute leukemia by identifying sites of resistance and potential relapse before they are clinically manifest, and monitoring patients with leukemic tumors to assure their eradication. Leukemic tumors should never be deemed “isolated” without radiologic documentation, as this study has shown that the majority of patients with one tumor will have multiple site involvement at some time. Improvement in prognosis for acute leukemia patients is long overdue, and an updating of our standard approach to treatment could decrease the high mortality from resistant leukemia in sites outside the marrow documented in this analysis.
Acknowledgements
The contributions and collegial spirit of the dedicated physicians who made this work possible are gratefully acknowledged, as are the invaluable suggestions of Drs. Gwen Nichols and David Straus
