Inhibition of the phosphoinositide 3-kinase pathway for the treatment of patients with metastatic metaplastic breast cancer

Molecular aberrations that can activate the PI3K/Akt/mTOR axis are common in mesenchymal/metaplastic breast cancer- and temsirolimus-based regimens offer promise for treatment of these rare cancers. Response rates (CR + PR) were 32% including 2 CRs in patients treated with liposomal doxorubicin and temsirolimus (±bevacizumab).

Abstract

Background

Mesenchymal/metaplastic breast cancers (MpBCs) are often triple-negative (TNBC), and chemo-refractory, and can harbor phosphoinositide 3-kinase (PI3kinase) alterations; thus, therapy with mTor inhibitors may demonstrate activity.

Patients and methods

Patients with mesenchymal/MpBC treated with temsirolimus-based regimens were evaluated. Mutational analyses [polymerase chain reaction (PCR)-based DNA sequencing method, mass spectrometric detection (Sequenom MassARRAY), or next-generation sequencing] as well as loss of phosphatase and tensin homolog (PTEN) (immunohistochemistry) were performed (archived tissue when available).

Results

Twenty-three patients (one of whom was on two separate trials) were treated using temsirolimus-containing regimens: temsirolimus alone (n = 1 patient) or combined with the following: liposomal doxorubicin and bevacizumab (DAT, n = 18); liposomal doxorubicin (DT, n = 1); paclitaxel and bevacizumab (TAT, n = 2); paclitaxel (TT, n = 1); carboplatin and bevacizumab (CAT, n = 1). Response rate [complete response (CR) + partial response (PR)] was 25% across all regimens; 32% in the anthracycline-based regimens [DAT and DT (CR = 2, PR = 4; N = 19)]. An additional two patients achieved stable disease (SD) ≥6 months [total SD ≥6 months/CR/PR = 8 (33%)]. Molecular aberrations in the PI3K pathway were common: PIK3CA mutation = 6/15 (40%), PTEN mutation = 3/11 (27%), and PTEN loss = 2/11 (18%). A point mutation in the NF2 gene (K159fs*16; NF2 alterations can activate mTor) was found in one patient who attained CR (3+ years). Of the eight patients who achieved SD ≥6 months/CR/PR, all 4 patients with available tissue had a molecular aberration that activate the PIK3CA/Akt/mTOR axis: PIK3CA mutation = 2; PTEN loss = 1; NF2 aberration = 1.

Conclusions

DAT has activity in MpBCs including complete CRs. Molecular aberrations that can activate the PI3 K/Akt/mTOR axis are common in MpBC.

introduction

MpBCs are defined as invasive carcinomas that have either squamous or mesenchymal differentiation and are commonly TNBC [1]. Despite variation in prognosis depending on grade and histologic variant, high-grade MpBCs are relatively chemo-refractory and carry a worse prognosis than nonmetaplastic, triple-negative breast cancers (TNBCs) [2–7].

Advances in molecular testing have further characterized TNBCs suggesting that subsets such as ‘claudin-low’, ‘mesenchymal’, or ‘mesenchymal-stem cell like’ are enriched in epithelial-to-mesenchymal transition (EMT), maintain stem cell-like characteristics, have PI3K pathway activation, and respond to PI3K inhibitors in preclinical models [8, 9]. Further characterization of claudin-low tumors revealed that 20%–30% are MpBCs [10]. Like claudin-low and mesenchymal/mesenchymal-stem cell-like tumors, MpBCs are also enriched in EMT, display PI3K pathway activation, and commonly carry PIK3CA mutations [11]. MpBCs also show strong correlation with a circulating stem cell-derived genomic profile heavily weighted for PI3K activity [11, 12]. Thus, treatment regimens that induce response in MpBCs may be used to treat the subsets of TNBC that lack metaplastic morphology, but share similar molecular characteristics. Given these molecular features, we report outcomes of patients with MpBC who were treated within the M.D. Anderson Cancer Center (MDACC) using regimens containing the mTOR inhibitor temsirolimus. Data from the first five patients treated have been previously published but were included to capture the entire experience for this rare cancer subtype [13].

patients and methods

patient selection and data collection

Patients who carried a diagnosis of MpBC (invasive myoepithelial carcinoma for one patient) confirmed by pathology review at MDACC were treated in accordance with MDACC internal review board (IRB) guidelines. Response assessments occurred every 6–12 weeks using Response Evaluation Criteria in Solid Tumors (RECIST) [14].

molecular correlative studies

When archived tissue was available, DNA was analyzed by polymerase chain reaction (PCR)-based DNA sequencing method for select genes of interest, particularly PIK3CA mutations [15, 16]. After January 2011, the assay was changed to mass spectrometric detection (Sequenom MassARRAY) [17]. In some patients, loss of PTEN was measured using immunohistochemistry (DAKO, Carpinteria, CA). Genes analyzed by PCR-based sequencing were not re-analyzed by Sequenom. Two patients' tumors underwent next-generation sequencing through Foundation Medicine (182 and 236 gene panel, Cambridge, MA).

results

patients and treatment

From May 2009 until September 2012, 23 patients were treated using six different temsirolimus-based therapy regimens (1 patient treated on two separate trials), yielding 24 analyzable outcomes (Table 1). All but one patient had metastatic cancer and this patient had T4, locally advanced cancer that was unresponsive to neoadjuvant anthracycline-based chemotherapy. All patient tumors were negative for ER/PR or HER2 with the majority (91%) having visceral disease. The most common histology was sarcomatoid carcinomas [including keratin-positive, pure sarcomatoid morphology (n = 7) or both sarcomatoid and epithelial components (n = 7)], followed by matrix-producing carcinomas (n = 7) Table 1. All of the matrix-producing carcinomas had chondroid or chondromyxoid stroma. Median number of prior chemotherapy regimens was 2 (range 0–7).

Table 1.

Patient characteristics (n = 23 patients)

Characteristics	No.	%
Age
Median	59
Range	21–86
ECOG
0	3	13
1	18	78
2	2	9
Metastatic disease
Yes	22	96
No	1	4
Visceral organ disease
Yes	21	91
No	2	12
Hormone receptor status
Positive	0	0
Negative	23	100
HER2 status
Positive	0	0
Negative	21	91
Unknown	2	9
Prior therapies for MBC
0	5	22
1	5	22
>1	13	56
Prior anthracycline
Yes	15	65
No	8	35
Prior bevacizumab
Yes	6	26
No	17	74
Predominant histology
Sarcomatoid, pure	7	30
Sarcomatoid, mixed	7	30
Matrix-producing	7	30
Other	2	9

All patients received therapy with the temsirolimus, most in combination with chemotherapy ± bevacizumab (Table 2). Liposomal doxorubicin was the usual chemotherapy backbone (n = 19), followed by paclitaxel (n = 3; including one patient previously treated with DAT) and carboplatin (n = 1). One patient received temsirolimus alone. Several patients (n = 6) who began therapy with liposomal doxorubicin lost access during a nationwide drug shortage and received targeted agents alone for ∼6 months. Two patients developed PD and discontinued therapy; one subsequently received therapy with PAT.

Table 2.

Response by treatment regimen

Treatment regimen	Dose	No. of treated	Median no. of cycles	CBR	RR
DAT	20 mg/m2, 15 mg/kg, 25 mg	5	6	2	2
	30 mg/m2, 15 mg/kg, 25 mg	13	5	4	3
PATa	60 mg/m2, 15 mg/kg, 25 mg	2a	7a	1a	0
CAT	4 AUC, 10 mg/kg, 25 mg	1	2
DT	30 mg/m2, 25 mg	1	12	1	1
PT	60 mg/m2, 25 mg	1	3
T	25 mg	1	2

^aOne patient previously treated with DAT and one patient treated neoadjuvantly.

D, liposomal doxorubicin; A, bevacizumab; T, temsirolimus; P, paclitaxel; C, carboplatin.

efficacy

The response rate [complete response (CR)/partial response (PR)] to PI3K/Akt/mTor pathway-directed therapy in treated patients (n = 24 outcomes) with MpBC was 25% (n = 6). All responding patients received liposomal doxorubicin for a RR of 32% to the DAT/DT regimens. One patient attained a CR by RECIST with DAT after eight cycles of therapy (two without bevacizumab due to toxicity) followed by temsirolimus alone for a year before transitioning to oral everolimus (10 mg daily), which she continues to date (4 years on) without evidence of disease (Figure 1). Notably, this patient was treated with DAT as first-line therapy for her metastatic disease and was anthracycline naïve as she had only received prior docetaxel and cyclophosphamide (five cycles) in the adjuvant setting. A second patient who achieved CR was also treated with DAT [progression free survival (PFS) = 22 months] (Table 4). This patient had received four cycles of doxorubicin and cyclophosphamide in the adjuvant setting and was treated with carboplatin and paclitaxel followed by radiation for a chest wall recurrence prior to beginning DAT for metastatic disease in the lung. Four patients achieved PRs; three were treated with DAT; and one with DT. Two additional patients receiving PI3K pathway-directed therapy in combination with chemotherapy achieved SD ≥6 months (DAT = 1, PAT = 1) for an overall clinical benefit rate (CBR = SD ≥6 months/CR/PR) of 35% (n = 8 of 23 patient outcomes) and 37% in those receiving DAT/DT.

Figure 1.

Scans of the chest showining response in patients treated with liposomal doxorubicin, bevacizumab and temsirolimus (DAT).

Table 4.

Characteristics of responders

Patient #	Response	Histology	Molecular	Treatment regimen	# of prior cytotoxic regimens (metastatic disease)	Prior anthracycline	# of cycles
01	CR	Sarcomatoid, pure	NF2 truncation; TP53 mutation	D 20 mg/m2; A 15 mg/kg; T 25 mg	1 (0)	None	8
09	CRa	Sarcomatoid, pure	PIK3CA mutation H1047R	D 30 mg/m2; A 15 mg/kg; T 25 mg	2 (0)	Adjuvant	17
02	PR (−40%)	Sarcomatoid, pure	Unknown	D 20 mg/m2; A 15 mg/kg; T 25 mg	1 (0)	None	4
12b	PR (−61%)	Matrix-producing	No loss of PTEN; mutation status unknown	D 30 mg/m2; A 15 mg/kg; T 25 mg	2 (0)	Neoadjuvant	5
17	PR (−32%)	Sarcomatoid, mixed	PTEN loss	D 20 mg/m2; A 15 mg/kg; T 25 mg	1 (0)	Neoadjuvant	15
18	PR (−40%)	Myoepithelial	Unknown	D 30 mg/m2; A 15 mg/kg; T 25 mg	1 (1)	None	22
06	SD ≥6 months (+8%)	Sarcomatoid, pure	PIK3CA mutation H1047R	D 20 mg/m2; A 15 mg/kg; T 25 mg	1 (0)	Adjuvant	7

^aPatient had PET avid disease prior to therapy with a minimal residual linear abnormality on CT chest that demonstrated no residual FDG uptake after 17 cycles of therapy; considered a CR. She transitioned to oral everolimus (10 mg daily) and later received consolidative stereotactic radiation to her isolated, treated pulmonary metastasis. PFS was 22 months (developed chest wall recurrence).

^bAlso experienced SD ≥6 months on PAT.

toxicity

Toxicities associated with DAT have been previously published and thus not detailed here [18]. However, it is important to note that one patient with a history of postobstructive pneumonia died of pneumonia during cycle 2 of therapy with DAT. A second patient treated with DAT had grade 3 congestive heart failure simultaneously with disease progression within the brain and chose to enter hospice, so it is unknown if her heart function improved after discontinuing therapy.

molecular correlative studies

Archived tissue was available for molecular analysis in 16 of 23 patients (one patient had breast recurrence and lung metastasis analyzed) (Tables 3–5). Half of the samples tested (n = 8) were from primary tumors and half (n = 8) were from recurrent disease (4 from local recurrences and 4 from distant metastatic disease). Nine samples were evaluated prior to 2011 using PCR-based sequencing and 7 were evaluated using mass spectrometric detection (Table 5). No genes analyzed by PCR-based sequencing were reanalyzed by mass spectrometric techniques; however, tumor samples from patient 15 underwent analysis of different genes, initially by PCR (PIK3CA, PTEN) and then by mass spectrometric techniques (BRAF, KRAS, AKT1). A molecular aberration in the PI3K pathway was detected in 69% of patients tested (Table 5). The majority harbored mutations in PIK3CA (6 of 15, 40%) or PTEN aberrations (5 of 15, 33%). Two of 11 (18%) had loss of PTEN by IHC and 3 of 11 (27%) harbored aberrations in PTEN (Table 5). Both patients who achieved CRs harbored molecular aberrations: one a PIK3CA mutation, the other an NF2 mutation (Table 5).

Table 3.

Response according to aberration status

Mutation (# tested)	# positive	CR	PR	SD ≥6 months	PD/SD ≤6 months
PI3 Kinase (n = 15)	6	1	0	1	4
PTEN mutation (n = 11)	3	0	0	0	3
PTEN loss by IHC (n = 11)	2	0	1	0	1
NF2 mutation (n = 2)	1	1	0	0	0
Unknowna or no aberration present (n = 7)	NA	0	2	1	4

^aInadequate tissue available.

Table 5.

Genomic aberrations and associated clinical features in patients with tissue available for analysis

Patient #	Histology	Regimen	Aberration	Genes or protein tested	Testing	Best response
1	Sarcomatoid, pure	DAT	None PTEN detected NF2 K159fs*16 TP53 F109fs*39	PIK3CA*, EGFR, BRAF, KRAS, PTEN PTEN expression 182 gene panel	PCR-based IHC Foundation Medicine	CR
3	Sarcomatoid, pure	DAT	None PTEN detected	PIK3CA*, EGFR, BRAF, KRAS, C-KIT PTEN expression	PCR-based IHC	SD (2 cycles)
5	Matrix producinga	DAT	PTEN (D92G)	PTEN, PIK3CA*, EGFR, BRAF, KRAS, C-KIT	PCR-based	SD (6 cycles)
6	Sarcomatoid, pure	DAT	PI3K (H1047R)	PIK3CA*, EGFR, BRAF, KRAS, C-KIT	PCR-based	SD ≥6 months (8 cycles)
7	Sarcomatoid, pure	DAT	PTEN (14 base pair deletion in exon 1)	PIK3CA**, PTEN	Sequenom	SD (4 cycles)
8	Sarcomatoid, mixed	DAT	PI3K (E545A and H1047Y) PTEN detected	PIK3CA* PTEN expression	PCR-based IHC	SD (4 cycles)
9	Sarcomatoid, pure	DAT	PI3K (H1047R) PTEN detected	PIK3CA*, BRAF, KRAS, C-KIT, TP53 PTEN expression	PCR-based IHC	CRb
10	Sarcomatoid, mixed	DAT	PI3K (H1047R) PTEN detected	PIK3CA*, BRAF, KRAS, PTEN PTEN expression	PCR-based IHC	NEc
12	Matrix producing	DAT and PAT	None PTEN detected	EGFR, KRAS, NRAS, C-KIT, AKT PTEN expression	Sequenom IHC	PR (DAT) then SD ≥6 months (PAT)
14	Sarcomatoid, mixed	DAT	PI3K (E545K) PTEN detected	PIK3CA*, EGFR, BRAF, KRAS, PTEN PTEN expression	PCR-based IHC	PD (2 cycles)
15	Sarcomatoid, mixed	T	None None PTEN detected	PIK3CA*, PTEN BRAF, KRAS, AKT1 PTEN expression	PCR-based Sequenom IHC	SD (2 cycles)
16	Matrix producing	DAT	None PTEN detected	PIK3CA**, EGFR, BRAF, KRAS, C-KIT, PTEN, MET, GNAQ PTEN expression	Sequenom IHC	PD (2 cycles)
17	Sarcomatoid, mixed	DT	No mutation PTEN loss	PIK3CA**, PTEN, MET PTEN expression	Sequenom IHC	PR (12 cycles)
20	Matrix producing	CAT	None	PIK3CA**, EGFR, BRAF, KRAS, C-KIT, PTEN, MET, GNAQ, GNAS, IDH1, IDH2, NRAS, RET	Sequenom	PD (2 cycles)
21	Sarcomatoid, pure	DAT	None	PIK3CA**, EGFR, BRAF, KRAS, C-KIT, PTEN, MET, GNAQ, GNAS, IDH1, IDH2, NRAS, RET	Sequenom	PD (2 cycles)
22	Matrix producing	DAT	PTEN loss PIK3R1 (N453_T454insNN) PTEN (K327fs*16) MCL1 amplification TP53 (N131del) KDM6A (Q692fs*37) RB1 (splice site 1421+1G > C)	PTEN expression 236 gene panel	IHC Foundation Medicine	PD (2 cycles)

Genes ‘evaluated by PCR’: EGFR (exons 18 to 21), BRAF (codons 595 to 600 in exon 15), KRAS (codons 12, 13 or 61), C-KIT (exons 9, 11, 13, 17) PTEN, exons 1–9, PIK3CA* (codons 532 to 554 in exon 9 or codons 1011 to 1062 in exon 20), PIK3CA**(codons 60, 88, 110, 111, 345, 405, 418, 420, 453, 539, 542, 545, 546, 909, 1021, 1025, 1043, 1046, 1047, or 1049), TP53 (exons 4 to 9), AKT1 (codons 17, 173, or 179), MET (codons 375, 848, 988, 1010, 1112, 1124, 1248, 1253, or 1268), GNAQ (codon 209), GNAS (codon 201), IDH1 (codon 132), IDH2 (codon 172), NRAS (codon 12, 13, or 61), RET (codon 918)

^aBy outside pathology review, slides not reviewed at M.D. Anderson.

^bPatient had PET avid disease prior to therapy with a minimal residual linear abnormality on CT chest that demonstrated no residual FDG uptake after 17 cycles of therapy; considered a CR. She transitioned to oral everolimus (10 mg daily) and later received consolidative stereotactic radiation to her isolated, treated pulmonary metastasis. PFS was 22 months (developed chest wall recurrence).

^cDeath during cycle 2 of therapy due to pneumonia, not evaluated for response (NE).

Of the four patients who attained PR, one had PTEN loss and the other three did not have adequate tissue available for complete testing (Table 3); of the two patients who achieved CR, both had aberrations that could lead to PI3K pathway activation (Tables 4 and 5). Overall, of the eight patients with clinical benefit, all four with correlative tissue had an aberration that could lead to activation of the PI3K/Akt/mTOR axis (Table 3). An additional patient with a detectable mutation in PIK3CA (H1047R) died of pneumonia during cycle 2 of therapy with DAT and was, thus, not assessed for response using RECIST; however, a reduction in size of her lung metastases was noted on chest X-ray prior to starting her second cycle of DAT. Not all patients with mutations responded. A patient with detected PIK3CA mutation E545K developed rapid disease progression after two cycles of DAT and the patient with detected PIK3CA mutations E545A and H1047Y developed disease progression after four cycles of DAT.

discussion

We describe the largest prospective series documenting treatment of MpBC, a rare but historically aggressive subtype of TNBC. MpBC includes a spectrum of histologic types, usually high grade and considered chemo-refractory. In retrospective series short-lived, partial responses have been noted with doxorubicin in patients with metastatic disease [5, 19]. Our prospectively collected data demonstrates that a variety of aberrations in the PI3K/Akt/mTor axis are a hallmark of these patients. Indeed, 10 of 16 patients with available tissue had an aberration in one or more of the following genes that affect this pathway (Table 5): PIK3CA, PTEN, NF2, or PIK3R1. Since next-generation sequencing was not available for many patients, it is plausible that other patients who had less comprehensive molecular testing might have also harbored an aberration in this pathway.

Additionally, we identified a CR/PR rate of 26% using chemotherapy with temsirolimus (±bevacizumab) and a 32% CR/PR rate with DAT/DT. Responses to DAT/DT included 2 CRs, one ongoing (with maintenance everolimus) 3+ years after discontinuing protocol therapy; time to progression from initiation of therapy in the second patient was 22 months.

Additionally, two other patients attained SD for ≥6 months for a CBR of 33% overall, 35% among those who received chemotherapy with temsirolimus (±bevacizumab) and 37% in those treated with DAT/DT. These response rates are consistent with rates published for PI3K-directed therapy in nonbreast cancer patients whose tumors harbor PIK3CA mutations [15, 16, 20, 21]. Most patients also received bevacizumab as a component of their therapeutic regimen. MpBCs commonly express VEGF and HIF-1, suggesting bevacizumab may be key to the success of the DAT regimen [22, 23].

The regimens were generally well tolerated with expected myelosuppression, mucositis, hand–foot syndrome, dyslipidemia and rarely pneumonitis [18]. One death occurred from postobstructive pneumonia and one patient had grade 3 congestive heart failure during DAT.

Though considered rare, MpBCs share similar molecular features to claudin-low and mesenchymal/mesenchymal-stem cell-like tumors, subsets comprising up to 30% of TNBCs [8, 9, 11]. In fact, MpBCs make-up a portion of tumors molecularly characterized as claudin-low [10]. Unlike other mesenchymal tumors, MpBCs can be identified using microscopy, thus have undergone more extensive molecular characterization. We have confirmed previously published rates of PIK3CA/PTEN mutations rates in MpBC and one CR tumor harbored a PIK3CA mutation. Additionally, a patient with durable CR (3+ years) to therapy exhibited NF2 mutation. NF2 encodes merlin, a negative regulator of the mammalian target of rapamycin complex 1 (mTORC1). Loss of function mutations in NF2 is associated with sensitivity to rapamycin in preclinical models of mesothelioma [24]. Interestingly, a durable complete response was seen with single-agent everolimus in bladder cancer with an NF2 nonsense mutation; further corroborating the potential for mTOR dependence in tumors with aberrations in NF2 [25]. Additionally, bevacizumab ± mTOR inhibition has reduced schwannomas, stabilized disease, and/or improved in symptoms in neurofibromatosis type 2 [26–28]. Trials are ongoing to determine the role of rapamycin analogs in neurofibromatosis [29, 30].

While the association of molecular aberrations with treatment response requires confirmation, three of six patients with CR or PR potentially activated the PI3K/Akt/mTOR axis (PIK3CA mutation, n = 1; PTEN loss, n = 1; NF2 aberration, n = 1; tissue unavailable in the other three). Both the patient with CR and another patient with SD ≥6 months had PIK3CA/H1047R mutation that is associated with response to mTOR inhibitors in early phase trials.

This study has important limitations. Notably, a range of regimens were employed, but each included temsirolimus, and all CRs and PRs occurred in regimens where liposomal doxorubicin (±bevacizumab) were included. Correlative data were also limited by tissue availability, although several molecular aberrations linked to activation of the PI3K/Akt/mTOR were identified. All patients with clinical benefit (CR/PR/SD ≥6months) and available tissue for molecular testing demonstrated these anomalies. Finally, comparative data consist of small, retrospective analyses involving diverse treatments.

In conclusion, patients with advanced MpBC harbor diverse molecular abnormalities potentially actionable by temsirolimus-based therapies which may achieve durable CR. These promising results justify exploration of the DAT regimen in a randomized clinical trial, including molecular correlates and nationwide recruitment of patients with this rare tumor subtype.

disclosure

RK receives consultant funds from Sequenom and is a founder of RScueRx, Inc. All remaining authors have declared no conflicts of interest.