Prospective Biomarkers for Chemotherapy Selection in Sarcoma

An ESUN Article by AD Schenone and BA Van Tine, MD, PhD

Introduction

Sarcomas are rare cancers of mesenchymal origin accounting for approximately 21% of all pediatric and 1% of all adult solid malignant tumors.¹ Within the United States in 2013 it was estimated there were 3,010 bone and 11,410 newly diagnosed soft tissue sarcomas (STS) with an estimated 1,440 and 4,390 deaths, respectively.^2,3 In addition to their rarity, sarcomas are highly heterogeneous with more than 70 histologic subtypes.⁴ Within each sarcoma subtype is a tumor heterogeneity that is further complicated by interpatient heterogeneity. Taken together this leads to the chemo-selectivity of an individual patient, which is why one patient will respond well to an agent and another with the same exact subtype does not.

Genetic and molecular techniques such as gene expression microarrays, fluorescent in situ hybridization (FISH), immunohistochemistry (IHC) and DNA sequencing are used by the biomarker field in order to identify patient outcomes (i.e. survival or response to treatment). For example, overexpression and amplification human epidermal growth factor receptor type 2 (HER2) is associated with response to Herceptin in HER2 positive breast cancer.⁵

There are many biomarker and DNA sequencing companies that have been marketing the use of various technologies for both clinical trial selection and chemotherapy prediction. Biomarkers in their simplicity are used quite commonly to make therapy decisions. For example, we readily use the protein expression ER, PR, and HER2 to treat breast cancer, EGFR mutation and ALK expression in lung cancer and EGFR and BRAF in colon cancer. The wider world of biomarker analysis is of utmost importance to the sarcoma field as we use classical chemotherapy as our mainstay. The main issue with biomarkers in sarcoma is their lack of validation either prospectively or retrospectively by our small field.

Therefore, we have put together a review of known biomarkers that exist for common therapeutics presently utilized in the treatment of sarcoma, not including gastrointenstinal stromal tumors (GIST). These include anthracyclines, taxanes, gemcitabine, dacarbazine and mTOR inhibitors. Unfortunately, evaluation of biomarkers for ifosfamide, cyclophosphamide, etoposide, and cisplatin treated patients is lacking across the literature. We have identified seven known biomarkers including TOP2A, RRM1, TLE3, MGMT, TUBB3, SPARC, and PTEN. Below, we have included known supporting evidence from unbiased and candidate searches of each biomarker along with a description of its function, associated chemotherapy, and relevance to sarcoma. It is our hope that these, as well as future DNA, RNA and protein biomarkers, will support appropriate treatment guidelines that can only be derived by appropriate correlative biomarker validation through prospective sarcoma clinical trials.

Biomarkers

1. Topoisomerase IIA (TOP2A)

Topoisomerases are enzymes that break and rejoin double stranded DNA. Type IIA topoisomerase (TOP2A) is essential to unwinding these strands during cell replication and repair (Figure 1). TOP2A cuts the DNA strands allowing them to uncoil and then reattaches each DNA strand. This process allows cells, including cancer cells, to replicate. In some cancers such as malignant peripheral nerve sheath tumors (MPNST), TOP2A can be highly active.⁶ Furthermore highly active TOP2A was found to be associated with poorer MPNST specific survival and presence of metastasis.⁶

Anthracyclines [Doxorubicin (Adriamycin®), liposomal-doxorubicin (Doxil®), epirubicin (Ellence®)] target cancer cells by attaching between base pairs of DNA/RNA strands, creating iron-mediated free oxygen radicals, removing histones from chromatin, and inhibiting topoisomerase IIA enzyme function. Most anthracyclines inhibit TOP2A by impairing the protein from reattaching DNA strands after the protein has cut a highly coiled region.⁷ This effectively freezes TOP2A and DNA strand inhibiting DNA repair by DNA ligase.⁸

The association between TOP2A and anthracycline activity has been seen in clinical trials. First, a meta-analysis by Du et al reviewed thirteen studies including 2,633 tumors from women receiving chemotherapy for locally advanced breast cancer.⁹ Patients with TOP2A amplified tumors receiving anthracycline-based neoadjuvant chemotherapy had a significant increase in relapse free survival (RFS) [hazard ratio (HR) = 0.64, 95% CI: 0.49-0.83, P = 0.001] and trends toward overall survival (OS) [HR = 0.59, 95% CI: 0.35-1.01, P = 0.056]. Second, a randomized control study analyzed 4,943 tumors from women with HER2-positive metastatic breast cancer receiving doxorubicin and cyclophosphamide with or without trastuzumab for TOP2A and HER2 amplification as measured by fluorescent in-situ hybridization (FISH).¹⁰ In patients with Her2/TOP2A-coamplified tumors receiving doxorubicin and cyclophosphamide without trastuzumab, there was a statistically significant increase in OS [38.5 months versus 18.2 months] and trends towards increased progression free survival (PFS) [7.1 vs. 5.6 months]. Finally, in the NCIC Clinical Trials Group Mammary 5 (MA.5) randomized control study 438 tumors from women with node-positive breast cancer who received cyclophosphamide, epirubicin and fluorouracil (CEF) versus cyclophosphamide, methotrexate and fluorouracil (CMF) were analyzed for TOP2A alterations (either amplifications or deletions) and HER2 amplification by FISH.¹¹ In TOP2A altered tumors, patients receiving CEF had a statistically significant increase in RFS [adjusted hazard ratio [HR] = 0.35, 95% confidence interval [CI] = 0.17 to 0.73, P = .005] and OS [adjusted HR = 0.33, 95% CI = 0.15 to 0.75, P = .008] versus CMF adjuvant chemotherapy. On a 2 year follow up, O'Malley and colleagues analyzed 477 tumors from 710 premenopausal women with node-positive breast cancer receiving CEF versus CMF adjuvant chemotherapy in the NCIC Clinical Trials Group Mammary 5 (MA.5) trial. 136 patients (28.5%) with TOP2A overexpressed tumors, as determined by immunohistochemistry, receiving CEF versus CMF adjuvant chemotherapy had a statistically significant increase in RFS (adjusted HR 0.45; 95% CI 0.25-0.82; P = 0.009) and OS (adjusted HR 0.50; 95% CI 0.26-0.96; P = 0.04).¹² However when tumors lacked TOP2A overexpression, CEF versus CMF adjuvant chemotherapy did not have a statistically significant increase in RFS [adjusted HR 0.88; 95% CI 0.64-1.22; P = 0.46] or OS [adjusted HR 0.95; 95% CI 0.66-1.38; P = 0.80].

Other restrospective studies have demonstrated a strong association between TOP2A and anthracyclines. In a retrospective biomarker analysis of the prospective randomized Scandinavian Breast Group trial 9401 analyzed 391 tumors from women with node positive breast cancer who received a tailored and dose-escalated epirubicin-based regimen versus epirubicin-based regimen followed by bone-marrow-supported high-dose cyclophosphamide, thiotepa, and carboplatin (CTCb) adjuvant chemotherapy.¹³ Patients with TOP2A-amplified tumors receiving tailored and dose-escalated epirubicin-based regimen versus epirubicin-based regimen followed by bone-marrow-supported high-dose CTCb had a statistically significant increase in RFS [HR 0.45; P = .049].

In a retrospective single-arm study, Rodrigo et al. analyzed tumors from 78 patients with locally advanced soft tissue sarcoma (LASTS) who received doxorubicin based neoadjuvant chemotherapy for ERCC1, HER2 and TOP2A expression by immunohistochemistry (IHC).¹⁴ Tumors with high TOP2A protein were associated with an improved rate of histological response and a significant improvement in DFS. Furthermore, tumors with low TOP2A protein had a significant decrease in median DFS. Most patients with high TOP2A protein tumors (73%) achieved a good histological response versus very few with low TOP2A protein tumors (33%). The median DFS in patients with low TOP2A protein tumors was 2.7 years and was not reached in patients with high TOP2A.

Breast cancers with high TOP2A amplified and overexpressed tumors are associated with improved clinical response to anthracycline containing regimens. However increased HER2 amplified and expressed breast cancers, both independently and in HER2/TOP2A co-amplified tumors, have been closely associated with an improved response to anthracycline based chemotherapy regimens. It is likely that this association may be related to the close proximity of their gene loci on chromosome 17q12-21. Preliminary, retrospective analysis of STS tumors with high TOP2A expression was also associated with improved clinical response. However this study represented a small number of patient tumors as comparable to studies performed in other cancers and may have been unduly influenced by a high prevalence of adjuvant radiotherapy.¹⁴ Additionally increased TOP2A expression has been identified in esophageal, gastric, colon, lung, and platinum refractory ovarian cancers.^15,16,17 Thus other cancer types, including sarcomas, warrant further investigation to determine if increased TOP2A amplified and overexpressed tumors are chemo sensitive to anthracycline containing regimens.

Relevance to Sarcoma: The main agents used to treat sarcoma are anthracycline based regimens. Unlike breast cancer, where there is a high cure rate with the use of chemotherapy, the response rate to anthracyclines in sarcoma has been defined prospectively by EORTC 62012, which demonstrated as 13.6% overall response rate and a 59.7 clinical benefit rate with a progression free survival of 4.6 months to doxorubicin.¹⁸ Though incorporation of biomarkers for anthracyclines has not become standard practice in carcinomas as it is likely not needed clinically, research has set precedence that TOP2A may be a possible biomarker. As this is a well-studied biomarker outside the sarcoma field, and it would benefit sarcoma patients if we can either validate or invalidate its use in sarcoma. Furthermore, if validated, TOP2A would allow us the opportunity to avoid these cardiotoxic regimens if they are predicted to fail.¹⁹

Although there has been extensive assessment of TOP2A amplification and expression, to our knowledge, further characterization of tumor specific TOP2A sequences including single nucleotide variants (SNV), and insertions/deletions (INDELs) have not been investigated against clinical outcomes. As specific TOP2A sequences variants may discern whether expression is causally implicated in therapeutic response, or a collateral biomarker reflective of an intact DNA damage response.

2. Ribonucleotide Reductase M1 unit (RRM1)

Ribonucleotide reductase is a key protein required for creating deoxyribonucleotides from ribonucleotides (Figure 2). Deoxyribonucleotides are the building blocks of DNA and are required for creating and repairing DNA strands. Ribonucleotide reductase (RNR) allows cells, including cancer cells, to form new deoxyribonucleotides. Ribonucleotide reductase subunit M1 (RRM1) is one of two components in the RNR protein. When RRM1 is inhibited cells, including cancer cells, lack sufficient deoxyribonucleotides to synthesize or repair DNA ultimately leading to cell death.²⁰

Nucleoside analogs are molecules that imitate nucleosides during DNA synthesis. As DNA is synthesized these analogs replace normal nucleosides. Some nucleoside analogs such as fluorouracil (5-FU) and gemcitabine (Gemzar®) target cancer cells. Gemcitabine targets cancer cells by replacing the nucleoside deoxycitidine during DNA synthesis and by inhibiting RNR function.²⁰ Gemcitabine inhibits RNR when it mistakes gemcitabine for a ribonucleotide disphosphate.²⁰ Once gemcitabine attaches, RNR is unable to remove it resulting in inhibition of its function.²⁰ Without new deoxynucleotides cells, including cancer cells, are unable to synthesize or repair DNA ultimately leading to cell death.²⁰ RNR consists of two components RRM1 and RRM2. Cancers cells with low levels of RRM1 are deficient in active RNR and are therefore more susceptible to its inhibition.

RRM1 and Nucleoside Analogs: In a meta-analysis eighteen studies including 1,243 non small cell lung cancer (NSCLC) tumors from patients with advanced NSCLC, with low or absent RRM1 versus high or present RRM1 amplified or expressed tumors receiving gemcitabine-containing regimens, had a significant increase in OS, 3.94 months longer (95% CI 2.15-5.73, P<0.0001) and PFS, 2.64 months longer (95% CI 0.39-4.89, P=0.02).²¹ NSCLC with low or absent RRM1 amplified or expressed tumors are associated with improved clinical response to gemcitabine containing regimens. RRM1's ability to produce deoxyribonucleotides is crucial to this response.

Relevance to Sarcoma: Gemcitabine and taxotere (GT) are one of the more common regiments used for the treatment of sarcoma. This regimen, though highly active in some subtypes of sarcoma, is less effective in others. Knowing the likelihood that either gemcitabine or taxotere will be active, either alone or in combination may allow for modification of this regimen minimizing toxicity. Thus other cancer types, including sarcomas, with low or absent RRM1 amplified or expressed tumors may be chemo sensitive to gemcitabine-containing regimens. Thus further investigation of GT in RRM1 low or absent sarcomas including validation by prospective clinical trials is warranted.

3. Transducin-like enhancer protein 3 (TLE3)

Transducin-like enhancer proteins are important proteins that help regulate cell-signaling pathways, including NOTCH and WNT pathways, known to be dysregulated among many cancers. Transducin-like enhancer protein 3 (TLE3) is known to help stop production of proteins from a number of mutated cancer forming genes which influence growth and microtubule stability.²²

Taxanes [Docetaxel (Taxotere®), paclitaxel (Taxol®), nab-paclitaxel (Abraxane®)] are molecules that impair microtubule function.²³ Microtubules are critical in cell division by organizing and separating chromosomes during mitosis (Figure 3). Taxanes inhibit removal of GDP-bound tubulin from the microtubule, thereby stopping microtubule function and freezing the cell in mitosis, ultimately leading to cell death.²³

TLE3 and Taxanes: In a retrospective well-designed biomarker analysis Kulkarni et al examined 411 tumors from women with early stage breast cancer to identify potential biomarkers associated with clinical response.²⁴ Presence of TLE3 was associated with an improved 5-year disease free interval (DFI) among all tumors (n = 441, P < 0.004). In patients with triple negative tumors (estrogen receptor (ER), progesterone receptor (PR) and Her2/neu), presence of TLE3 had a significant increase 5-year DFI among all triple negative tumors (n = 81, P < 0.015), and among patients receiving doxorubicin and cyclophosphamide (AC) + taxanes (T) (n = 45, P < 0.02). However in patients with triple negative tumors, presence of TLE3 5-year DFI was not significantly increased in patients receiving AC alone (n = 17, P = 0.81). In addition, a retrospective single-arm study by Ross et al analyzed 368 tumors from patients with NSCLC who received adjuvant taxane containing regimens versus surgical resection and/or radiation alone for presence of TLE3 by immunohistochemistry staining.²⁵ Patients with high TLE3 expressing tumors receiving adjuvant taxane containing regimens had a significant decrease in local and distant recurrences (HR=0.42 p=0.0047 n=68). However, high TLE3 expressing tumors receiving surgical resection and/or radiation alone did not have a significant decrease in local and distant recurrences (p=0.83 n=300). Finally, in retrospective analysis of breast and NSCLC with present or high TLE3 tumors, there is an associated improved clinical response to taxane containing regimens. TLE3 is found to regulate cancer-forming genes that influence growth and microtubule stability.²⁵

Relevance to Sarcoma: Along with gemcitibine, determining a biomarker for taxane response, such as TLE3, would aid the sarcoma field to determine which tumors are most likely chemo sensitive to taxane containing combination regimens, as single agent taxanes have historically proven largely ineffective with the exception of angiosarcoma with a response rate of approximately 20%.²⁶ By validating a biomarker for this class of drug we could reduce adverse events by taxanes including unnecessary neuropathy by determining if high TLE3 sarcomas are chemo sensitive to taxane and taxane containing regimens. Thus further investigation of taxane based regimens in TLE3 present or high tumors including validation by prospective clinical trials is warranted.

4. O6-methylguanine-DNA-methyltransferase (MGMT)

O6-methylguanine-DNA-methyltransferase (MGMT) is a protein that repairs naturally occurring methyl damage to the DNA nucleotide guanine.²⁷ MGMT acts by transferring the methyl damage to a cysteine nucleotide before other DNA repairing systems can replace it with an incorrect nucleotide (Figure 4).²⁰ When MGMT activity is low it leads to an increasing number of unrepaired methylated guanine nucleotides that results in DNA mutations and an increased risk of cancer.²⁷ Since cancer cells generally grow and divide faster than regular cells they typically have less time for DNA repair, increasing their sensitivity to methylated guanine nucleotides.²⁷ Certain chemotherapy agents have alkylating properties that take advantage of this sensitivity by adding methyl groups to nucleotides.²⁷

Alkylating agents [Temozolomide (Temodar®), Dacarbazine (DTIC®)], methylate guanine nucleotides resulting in DNA double helix strand cross-linking. These cross linkages inhibit DNA repair and synthesis ultimately leading to cell death. MGMT repairs methylated guanine such as those created by temozolomide and dacarbazine.

MGMT and Alkylating Agents: In a single-arm phase II study Chinot, O.L. et al analyzed 25 tumors from patients with inoperable newly diagnosed glioblastoma receiving neoadjuvant temozolomide prior to radiotherapy was analyzed for MGMT expression.²⁸ Patients with low MGMT versus high MGMT expressing tumors, had a significant increase in response rate (55%, n=11 vs. 7%, n=14; P = .004), PFS (5.5 vs. 1.9 months; P = .009) and OS (16 vs. 5 months; P=0.003). Additionally, a retrospective biomarker analysis of the prospective-observational study by Spiegl-Kreinecker et al analyzed 71 tumors from patients with glioblastoma multiforme (GBM) receiving or not receiving temozolomide for MGMT promoter methylation by- methylation-specific PCR and MGMT protein expression by western blot finding 47 of 71 (66%) tumors having methylated MGMT gene promoters, and 37 of 71 (52%) having MGMT protein expression, respectively.²⁹ MGMT protein expressing tumors, but not promoter methylated tumors, and temozolomide therapy had a significant association (P = .015) with MGMT protein expression having a significant decrease in OS (HR 5.53, 95% CI 1.76-17.37; P = .003). While patients not receiving temozolomide with MGMT protein expression had no significant difference in OS (HR 1.00, 95% CI 0.45-2.20; P = .99).

In a retrospective analysis of a previously conducted prospective trial Ma, S. et al analyzed lymph node metastasis in 65 patients with metastatic melanoma prior to receiving single agent dacarbazine (DTIC) or dacarbazine containing regimens for MGMT protein expression.³⁰ 12 patients had more than one lymph node analyzed with 7 out of 12 having variable MGMT expression. Six out of 12 (50%) responding versus 12 out of 53 (23%) non-responding patients had less than 50% MGMT positive tumor cells (P = 0.077). In a Retrospective single-arm study, Levin et al. analyzed 28 tumors from patients with low grade oligodendrogliomas without prior radiotherapy receiving temozolomide for MGMT protein expression by IHC.³¹ Low/intermediate (5 patients) versus high (4 patients) MGMT expression had a significant increased objective response (P < .047). Finally, a retrospective case control study Kulke, M.H. et al. analyzed 97 tumors from patients with neuroendocrine tumors receiving a temozolomide-based regimen for MGMT expression by IHC. 19 out of 37 (51%) pancreatic neuroendocrine and 0 out of 60 (0%) carcinoid tumors were MGMT deficient.³² From 21 patients with analyzable tumor, 4 out of 5 (80%) MGMT deficient and 0 out of 16 (0%) MGMT expressing tumors responded to temozolomide (P = 0.001).

Glioblastoma, low grade oligodendrogliomas, melanoma and pancreatic neuroendocrine tumors with low MGMT amplified or expressed tumors are associated with improved response to alkylating containing regimens. High MGMT amplified or expressed tumors are associated with reduced response to alkylating containing regimens. MGMT role transferring methyl damage from alkylating containing regimens is crucial to this response.

Relevance to Sarcoma: Although tumor sample sizes are small, preliminary results indicate that the use of DTIC and dacarbazine is associated with a low but real response rate in sarcoma.³³ Identification of patients with a high likelihood to either respond or not respond to these agents is necessary. We currently have two phase III clinical trials using either Eribulin [NCT01327885]or Yondellis [NCT01343277] that have recently completed enrollment. The control dacarbazine arms of these trials allow the opportunity for validation of MGMT methylation as a prognostic marker in STS sarcoma. Furthermore, there are many agents with alkylating properties, such as ifosfamide and cyclophosphamide, presently utilized in the treatment of sarcoma that warrant clinical investigation to determine if low MGMT expressing tumors are associated with improved responses.

5. Tubulin beta-3 chain (TUBB3)

Tubulin is a protein that makes up the majority of the microtubule, a key component within the internal cellular structure and transportation system. Tubulin beta-3 chain (TUBB3) is a tubulin protein that helps maintain this internal cellular framework during cell division.

Taxanes/Vinca Alkaloids: As discussed prior, Taxanes [docetaxel (Taxotere®), paclitaxel (Taxol®), nab-paclitaxel (Abraxane®)] are molecules that impair microtubule function by inhibiting removal of GDP-bound tubulin (Figure 5) thereby stopping microtubule function ultimately leading to cell death. Vinca alkaloids [vinblastine (Velbe®), vincristine Oncovin®, vindesine (Eldisine®), vinorelbine Navelbine®] are molecules that impair microtubule function produced by the periwinkle plant catharanthus roseus. Microtubules are critical in cell division by organizing and separating chromosomes during mitosis. Vinca alkaloids inhibit addition of GTP-bound tubulin to the microtubule (Figure 5) thereby stopping microtubule function and freezing the cell in mitosis, ultimately leading to cell death. Tubulin beta-3 chain (TUBB3) is a tubulin protein that helps maintain microtubules during cell division.

TUBB3 and Taxanes/Vinca Alkaloids: In a single-arm study Azuma, K. et al analyzed 45 tumors in patients with relapsed NSCLC receiving carboplatin and paclitaxel for TUBB3 expression by IHC.³⁴ Negative TUBB3 expressed tumors had a significant increase in median PFS (40 weeks vs. 35 weeks, P=0.031) and a numerical, but not a statistically significant increase in OS (78 weeks vs. 57 weeks, P=0.087). While in a retrospective biomarker analysis of a randomized controlled study Vilmar, A.C. et al. analyzes 443 tumors in patients with advanced NSCLC receiving vinorelbine- or paclitaxel-containing chemotherapy for TUBB3 expression by IHC. 261 out of 443 (58.9%) patients had sufficient tumor sample for TUBB3 evaluation.³⁵ Low versus high TUBB3 expressing adenocarcinoma tumors had a significant increase in median PFS (7.87 versus 6.83 months P = 0.035) and median overall survival (14.17 versus 11.17 months P = 0.018).

NSCLC cancer with negative or low TUBB3 expressed tumors are associated with improved clinical response with taxane and vinca alkaloid containing regimens. TUBB3's role maintaining microtubule structure during cell division is common among many cells. Thus other cancer types, including sarcomas, with negative or low TUBB3 expressed tumors may be chemo sensitive to taxane and vinca alkaloid containing regimens.

Relevance to Sarcoma: The low response rate to vinorelbine in sarcoma makes this one of the more rarely used forms of therapy in sarcoma. The combination of vinorelbine with gemcitabine has been studied in sarcoma and demonstrates a response rate of approximately 13%.³⁶ If a validated biomarker for vinorelbine were to be established, such as TUBB3, this may become a more useful chemotherapeutic in the armamentarium of therapies against sarcomas.

6. Secreted protein acidic and rich in cysteine (SPARC)

Secreted protein acidic and rich in cysteine (SPARC) is an extracellular glycoprotein involved in bone and blood vessel formation as well as cell growth and migration. High levels of SPARC have been found in many cancers including some sarcomas.³⁷ SPARC was also found to preferentially attach to serum albumin, a common protein found in the blood.³⁸

As discussed prior, Taxanes [docetaxel (Taxotere®), paclitaxel (Taxol®), nab-paclitaxel (Abraxane®)] are molecules that impair microtubule function by inhibiting removal of GDP-bound tubulin thereby stopping microtubule function ultimately leading to cell death. Chemotherapies such as Nab-paclitaxel are an albumin coated taxane (Figure 6).

SPARC and nab-paclitaxel: In a retrospective biomarker analysis Morgan et al. analyzed 27 tumors from patients with a range of undetermined STS subtypes for SPARC expression by IHC. Moderate to elevated (15/27, 56%) versus low to moderate (12/27, 44%) SPARC expressed tumors had a significant decrease in median survival (4.4 months versus 22.1 months; P = 0.0016).³⁷ Despite limited sample size and an indeterminate STS subtype distribution, these preliminary findings warrant further investigation to establish SPARC expression among STS subtypes and to determine whether high SPARC expressing tumors are chemo sensitive to albumin coated chemotherapy based regimens such as nab-paclitaxel.

Relevance to Sarcoma: This early work deserves further attention in the field. Carefully incorporating biomarkers into the clinical trials may help identify patient for whom gemcitabine nab-paclitaxol may be superior to gem taxotere.

7. Phosphatidylinositol-3,4,5-trisphosphate 3-phosphatase (PTEN)

Phosphatidylinositol-3,4,5-trisphosphate 3-phosphatase (PTEN) is a protein that removes one phosphate group from phosphatidylinositol-3,4,5-trisphosphate (PIP3) forming its deactivated phosphatidylinositol 4,5-bisphosphate (PIP2) form. PTEN has a number of known internal cellular signaling roles including regulation of the PI3K/AKT pathway (Figure 7). When this pathway was left unregulated by removal of the PTEN gene in mice they developed leiomyosarcomas, a subtype of STS.³⁹

mTOR is a serine/threonine kinase that plays a critical role in cell growth, proliferation, angiogenesis, and survival. PTEN has a number of roles including regulation of mTOR pathways by deactivating PIP3. First generation mTOR inhibitors [Sirolimus (Rapamune®), Temsirolimus (Torisel®), Everolimus (Afinitor®), Ridaforolimus (AP23573 and MK-8669)] attach to mTOR intracellular receptor FKBP 12 stopping MTOR complex 1 (MTORC1) formation and activation.⁴⁰

PTEN and mTOR Inhibitors: In a retrospective biomarker analysis Filip Janku et al. analyzed 461 tumors from patients with a diverse group of cancers including breast (15), colorectal (82), pancreatobiliary (24), prostate cancers (19), renal (11), salivary gland (10), and uterine (33) receiving targeted therapy for MAPK and PTEN expression by IHC. 293 out of 461 (64%) had negative or reduced PTEN expressing tumors.⁴¹ 17 out of 153 (11%) patients treated with PI3K/mTOR inhibitors had partial responses including 2/13 (15%) negative, 12/90 (13%) reduced and 3/50 (6%) positive PTEN expressing tumors. No significant increase in partial responders among positive versus negative or reduced PTEN expressing tumors was observed (3/50, 6% vs. 14/103, 14%; p=0.27). However negative or reduced versus positive PTEN tumors had a significant increase in MAPK expressing tumors (72/144, 50% vs. 20/62, 32%, p=0.02). Furthermore, a retrospective biomarker analysis by Filip Janku et al. analyzed 1,656 tumors from patients with advanced refractory cancers receiving various treatments including targeted agents for phosphatidylinositol-4,5-bisphosphate 3-kinase, catalytic subunit alpha (PIK3CA) and PTEN genetic alterations.⁴² 146 out of 1,589 (9%) PIK3CA mutated and 149 out of 1,157 (13%) PTEN loss or mutated tumors were identified. In multicovariate analysis 136 out of 203 PIK3CA and/or PTEN genetic altered tumors in patients receiving PI3K/AKT/mTOR inhibitors had a significant increase in partial response (odds ratio = OR 4.34, 95% CI 1.23–15.24; p = 0.02). Furthermore, among this group only combination therapies had a significant increase in partial response (PR) (OR 5.31, 95% CI 1.16–24.25; p = 0.03) and stable disease for greater than six months/PR (OR 4.99, 95% CI 1.39–17.89; p = 0.01). However, in one subset of PIK3CA H1047R mutated versus other PIK3CA mutated tumors in patients receiving PI3K/AKT/mTOR inhibitors had a significant increase in PR (35% versus 12%; p = 0.039), SD for greater than six months/PR (45% versus 17%; p = 0.016), and PFS (4.6 months versus 2 months; p = 0.03).

Negative or reduced PTEN amplified or expressing tumors are found among a broad and diverse set of cancers, including sarcomas. Recently the SUCCEED trial, a randomized control phase III trial of patients with metastatic STS or bone sarcoma receiving ridaforolimus found a significant improvement in PFS. However, significant heterogeneity among PTEN altered tumors including MAPK and PIK3CA alterations may be confounding prior investigations. Thus new clinical investigations including retrospective biomarker analysis of PTEN altered tumors for MAPK and PIK3CA alterations may identify PI3K/AKT/mTOR inhibitor sensitive tumors.

Relevance to Sarcoma: Though mTOR inhibition is not a classic chemotherapy intervention, the loss of PTEN as demonstrated at ASCO 2014 Movva et al.,⁴³ is a common feature of sarcoma which should be further exploited and studied in trial. Determining the subset of sarcomas that will respond to mTOR inhibition is vital as these oral medications are well tolerated. Thus further investigation of PTEN altered tumors including MAPK and PIK3CA altered sarcomas including validation by prospective clinical trials are warranted.

Summary

We undertook this review to identify biomarkers associated with chemotherapeutic agents presently utilized in the treatment of sarcoma. We have reviewed the evidence associating increased TOP2A, and TLE3 with response to anthracycline and taxane containing regimens, respectively. Simultaneously, we discussed the available biomarker evidence associated with decreased RRM1, MGMT, and TUBB3 expression and benefit from gemcitabine, alkylating and taxane/vinca alkaloid containing regimens, respectively. Furthermore, new data in sarcoma may associate increased SPARC levels with benefit from albumin coated chemotherapy based regiment nab-paclitaxel. Additionally, further clinical investigation of agents with alkylating properties, such as ifosfamide and cyclophosphamide, or topoisomerase inhibiting properties, such as etoposide, is warranted to determine whether these agents are associated with increased response to low MGMT and high TOP2A expressing tumors, respectively. Finally, new clinical investigations including retrospective biomarker analysis of PTEN altered tumors for MAPK and PIK3CA alterations may identify PI3K/AKT/mTOR inhibitor sensitive tumors.

Conclusion

We identified TOP2A, RRM1, TLE3, MGMT, TUBB3, SPARC and PTEN as biomarkers associated with chemotherapeutic agents that need to be validated for the treatment of sarcoma. It is our hope that these, as well as future biomarkers, will be utilized in clinical decision making for appropriate treatment and prospective trail design. Biomarkers are powerful tools, both in their positive and negative predictive value. Knowing what agents are less likely to work is as important as knowing what agents are more likely to work. Therefore, by validating biomarkers through sarcoma specific clinical trails, we may one day be able to select the agent or combination of chemotherapy agents most appropriate for every sarcoma patient.

Biomarkers have many limitations to the sarcoma field. Presently we are extrapolating data from other tumor types and making the assumption that this translates to the mesenchymal tumors that we treat. The Bisgrove study demonstrated that biomarkers will have approximately a ~30% positive predictive value (PPV) in a heavily pretreated patient population.⁴⁴ On the other hand, the negative predictive value (NPV) of biomarkers will always be higher as the absence of the expression of a biomarker that positively correlates with outcomes (for examples, ER and Tamoxifen), has a much higher likely hood of predicting failure.

Many of our Phase II and III clinical trials collected tissue and have outcomes data. At this time the most important outcome biomarker to assess should be TOP2A and doxorubicin. The PICASSO clinical trial has outcomes data and tissue collected as well as the Threshold Clinical Trial. These both have doxorubicin alone arms with a large number of patients. As a community, we should be assessing the chemotherapy outcomes of the control arms of these studies against TOP2A. Other planned phase II and III trials that are currently enrolling or are being planned should incorporate arms that allow us to validate the markers described above. The positive and negative predictive values once defined will allow the sarcoma oncologists data that can be used to make treatment decisions. Sarcoma is a chemo-selective disease and biomarkers may be useful in making this selection, but only if the sarcoma community insists that the large trials that we are performing define the biomarkers that we need. In addition, we must not forget that the negative predictive value (NPV) of a biomarker will always be stronger than the positive predictive value. Once defined, the NPV could be used to bypass therapies that are most likely to fail which will avoid toxicity that comes with a low likelihood of benefit.

Conflict of Interests: Brian Van Tine is a consultant and on the speakers bureau of Caris Life Sciences.