Orchestrating the Tumor Microenvironment to Improve Survival for Patients With Pancreatic Cancer Normalization, Not Destruction

Hyaluronan

Hyaluronan is the protein-free polysaccharide found in many tissues throughout the body, with higher concentrations found in the cartilage, skin, eyes, and especially in synovial fluid. Hyaluronan binds multiple water molecules, enhancing volume, rigidity, and viscosity of fluids. In normal tissues, HA is maintained through a balance of HA synthase (HAS) and hyaluronidase activities. Considering its main cellular functions, many had long thought that HA was molecularly inert. With new understanding of the role of the hyaladherins, as well as the HA-binding receptors, CD44, and the receptor for HA-mediated motility (RHAMM), it is now clear that HA plays a much more direct role in regulating many biological processes including cell adhesion, proliferation, and migration, as well as cell invasion and recruitment of neutrophils into the tumor bed.

Hyaluronan levels appear to be elevated in many cancers, including in PDAC, and it often localizes to the stromal compartment of tumor tissues. Hyaluronan has been shown to be involved in the invasion of PDAC cells in experiments demonstrating HA synthesis at the invasive front of expanding tumors.²³ Bertrand and colleagues²³ report a more than 4-fold increase in HA deposition at the invasive tumor front relative to the adjacent normal tissues. Hyaluronan binding to CD44 promotes tumor migration of colon, breast, and brain cancer cells. In addition, it appears that the level of HA correlates with tumor migration and proliferation indices in breast or ovarian cancer. Induction of expression of the HAS, HAS1, also increases cellular proliferation rates. While studies are ongoing to determine the utility of HAS1 as a biomarker, HA levels may prove to be a suitable biomarker in several cancers including bladder cancer. As HA and HAS1 levels appear altered in multiple disease states, it seems that a balance of HAS and hyaluronidase activities is necessary to maintain the normal tissue homeostasis and function of HA.

Elevated levels of HA have been observed in several malignancies including pancreatic and gastric cancers, as well as in breast and bladder cancers.^24–27 In a recent report, we demonstrated that HA negatively correlates with the survival of patient with PDAC.¹⁷

Prior studies revealed that HA synthesis can be regulated by corticosteroid administration. For example, cortisol reduced the expression of HA by HAS by 50% in smooth muscle cells. Indeed, Gebhardt and colleagues²⁸ report that HA levels decreased by 50% following topical administration of the corticosteroid dexamethasone. Unfortunately, research of the effects of corticosteroid treatment on HA deposition in solid tumors is limited. Further study may be needed to fully understand its capabilities in solid tumor treatments.

Additional agents that have been reported to inhibit the production of HA by HAS include 4-methylumbelliferone (MU). 4-Methylumbelliferone reportedly synergizes well with gemcitabine in cancer cell lines, with little activity as a single agent. One interesting observation made by Yoshihara and colleagues²⁹ was that MU could decrease the incidence of metastasis in in vivo models for melanoma. 4-Methylumbelliferone also shows activity against squamous cell carcinoma, and prostate and esophageal cancer cell lines, reducing proliferation and migration. 4-Methylumbelliferone (also known as hymecromone) is used clinically as a choleretic and antispasmodic and is sold under multiple commercial names in countries throughout the world (e.g., Adesin C in Japan). It has entered clinical trials in the United States for chronic hepatitis infections (clinicaltrials.gov, NCT00225537), but results from its phase II trial have gone unreported and are unclear. Yet MU is the subject of multiple recent publications and represents an active area of continuing research.^30,31

While inhibiting its synthesis has been one approach, promoting breakdown of HA has also been an area of active research too. The hyaluronidases, which include 6 gene isoforms including HYAL1, HYAL2, HYAL3, HYAL4, PHYAL1, and PH20, balance HA levels and overproduction by catabolizing HA to monosaccharides and disaccharides, and smaller HA fragments, acting much as endo-β-acetyl-hexosaminidases. HYAL1 and HYAL2 are the most active, catabolizing as much as 30% of total HA each day in mammals. Hyaluronidase treatment has been proposed as a means to enhance degradation of HA overproduction and accumulation. Administration of hyaluronidase has been shown to synergize well with chemotherapeutics, in multiple tumor types, including pancreas, breast, and brain tumors, as well as in melanoma and sarcoma. The synergy is thought to arise from enhanced penetration of drugs gained following reductions of intratumoral pressures brought on by reduction in HA levels. In some models, the HA degradation may affect the signaling mediated by HA receptors such as CD44.

In preclinical models, hyaluronidase appears to decrease innate chemoresistance in spheroid culture tumor models, enhancing drug penetration and efficacy.³² In xenograft models, additional reports demonstrate enhanced doxorubicin penetration in hyaluronidase-treated tumors.³³ These improvements in penetration can exceed 15-fold enhancement in intratumoral concentrations of drugs such as melphalan.³⁴ These results are particularly encouraging as enhancing intratumoral drug concentrations is a continual battle in the effort to maximize both the efficacy and pharmacokinetic properties of drugs. This effort represents an approach, independent of any particular drugs' properties, which may enhance overall efficacy in a variety of conditions in patients.

Early pilot clinical trials used bovine hyaluronidase in efforts to modulate HA levels in patients. Some of these clinical trials are reviewed in Baumgartner's³⁵ report. One recurring observation was that despite its limitations, hyaluronidase enhanced efficacy in a variety of trial designs. One study demonstrated enhanced activity in the combination of vinblastine and hyaluronidase in Kaposi sarcoma, resulting in reduced recurrence when compared with vinblastine alone.³⁶ Reduced recurrence was also reported by another study looking at carboplatin and etoposide treatment in brain malignancy.³⁷ With a short infusion of hyaluronidase, there was a significant improvement of event-free survival and overall survival. Lastly, this synergism was also seen in patients with bladder cancer where intravesical administration of bovine hyaluronidase reduced recurrence from 32% to 7% with the addition of hyaluronidase (P < 0.05).³⁸

One particular limitation of bovine hyaluronidase has been the development of allergic reactions to this enzyme of bovine origin. As many as 32% of patients were reported to be preimmune to bovine hyaluronidase formulations, resulting in various inflammatory responses, including anaphylaxis.³⁵ Patients who do not already possess cross-reacting antibodies prior to treatment quickly develop antibodies that may preclude any chronic usage of a bovine hyaluronidase-containing regimen.

A recombinant, human hyaluronidase was recently developed (Hylenex) to alleviate some of the major shortcomings of bovine hyaluronidase. Some of the early clinical studies demonstrated that recombinant hyaluronidase did not induce allergic reactions in healthy volunteers following intradermal administration.³⁹ Hylenex is now Food and Drug Administration approved for improved dispersion and delivery of various injectable drugs.³⁹ Recombinant hyaluronidase is currently in clinical trials for multiple indications, including a partially randomized phase I/II trial in patients with newly diagnosed metastatic pancreatic cancer (clinicaltrials.gov, NCT01959139). As HA is synthesized in many locations in the body and used in many normal functions, development of adverse effects such as joint pain has been of particular concern. One report had suggested that inflammation or pain in joints was controlled and alleviated with administration of corticosteroids.³⁹ In the completed clinical trial of single-agent pegylated recombinant hyaluronidase, PEGPH20 (at 50 μg/kg), there was grade 3 muscle/joint pain, whereas doses of 0.5 and 0.75 μg/kg appeared to be well tolerated.⁴⁰

Because of its elevated levels and its role in desmoplasia in PDAC, HA is currently being pursued as a drug target in clinical trials in patients with advanced PDAC (clinicaltrials.gov, NCT01839487). The initial results of a randomized clinical trial of nanoparticle paclitaxel + gemcitabine, with or without pegylated hyaluronidase PEGPH20, have just been reported, and tumors with high hyaluronic acid levels appear to confer a longer survival after treatment with the triple-drug regimen. The overall results of that important trial are still pending.⁴¹

There has been considerable controversy as to whether compression of the tumor vasculature is caused by HA-induced fluid pressure or by solid tissue stress.^15,42–44 However, it is fair to say that HA is a reasonable candidate for therapeutic intervention.

Collagen

Collagens I, III, and IV are expressed and deposited at high levels in PDAC. As a major component of desmoplasia, collagens too have been pursued as a potential target in the development of stromal targeting agents (Table 1).^45–53 Tumor collagen levels have been shown to correlate negatively with macromolecule penetration.⁵⁴ In U87 (glioblastoma) and LS174T (colon adenocarcinoma) xenograft tumor model studies, Netti and colleagues¹⁰ demonstrated that high collagen levels were associated with significantly decreased diffusion of macromolecules, such as immunoglobulin G antibodies, resulting in 2.1-fold decreases in diffusion coefficients in high versus low collagen content models. Importantly, when collagenases were added to the model system, the decrease in diffusion in high collagen content tumors was abrogated. This and other reports found that collagen content directly impacts diffusion rates and suggested that other stromal macromolecules may also be perturbed in high collagen content tumors.

Collagens can act functionally through interaction with integrins. Integrin subunit expression has been shown to be dysregulated in some cancers. Integrin receptors, such as α_vβ₃, α_vβ₆, and α₆β₄, are thought to be involved in cancer progression.^55–57 In addition to signaling via integrin association in cancer, collagens can function in a similar fashion to HA by decreasing penetration of macromolecules into tumor tissues, presumably resulting in poor drug penetration and patient outcomes in PDAC.

In a prior report, Erkan and colleagues⁵⁸ reported that there existed a positive correlation between patient survival and collagen deposition. The difference in observations made by Erkan et al⁵⁸ and our report, which found that high collagen levels in patients' tumor were associated with poor patient survival,¹⁷ might be explained by their use of the aniline blue dye, which captures all collagens and our use of the collagen- or HA-specific antibodies. It may even suggest that loss of specific collagen subtypes may account for the overall reduction in collagen level Erkan and colleagues found associated with the poor survival cohort. Another explanation for difference in findings might be found in the treatments that the patients received that may have altered the collagen deposition that was observed at the time of resection.

Our more recent findings¹⁷ indeed suggest that surrogate markers of desmoplasia are observed in metastatic lesions of PDAC. Considering multiple sites of metastasis, it appears that both primary tumors and metastatic lesions show comparable levels of desmoplasia, including high levels of collagens I, III, and IV. Our data from patient-matched primary and metastasis samples are consistent with the notion that high levels of desmoplasia are observed in both primary tumors and metastatic lesions. Acquisition of pretreatment and posttreatment tissue samples or biopsies would allow an analysis of the effects of treatment on the deposition of ECMs in the course of chemotherapeutic intervention. As we looked only at excised tissue samples, our data reflected most directly the retention of ECMs following treatment. However, our data, taken together with studies demonstrating a negative correlation between ECMs such as collagen and the penetration of macromolecules, suggest that patients with fibrotic primary tumors and metastatic lesions would benefit from therapies targeting the tumor stroma.⁵⁴ The cellular lineage of the myofibroblasts giving rise to desmoplasia in PDAC is not entirely clear, although it has been suggested that they arise from the primary tumor.²² Certainly, in the liver, where PDAC commonly metastasizes, desmoplasia is also a recognized feature of tumors arising spontaneously. Our data suggest that the noncell autonomous tumor-signaling mechanisms that give rise to desmoplasia appear to be retained postmetastasis. A recent study by Suetsugu et al⁵⁹ demonstrates that stellate cells could accompany pancreatic cancer cells to distant metastatic sites.

Besides ECM deposition and desmoplasia, collagen has also been shown to play an important role in EMT and metastasis (Table 1). For instance, Shields et al^47,60 reported that collagen induces the expression of Snail gene, which in turn accelerates EMT. In addition, Collagen I was reported to support the maintenance of ALDH⁺ pancreatic cancer stem cells (Table 1).⁵²

There have been a few attempts to decrease collagen production. These have included losartan, relaxin, and candesartan (Table 1).

TGF-β Signaling

Increased TGF-β expression has been observed in multiple cancers, including breast, lung, hepatocellular, and pancreatic cancer. Importantly, TGF-β expression has been correlated with tumor metastasis and progression and poor outcomes in patients.⁶¹ In prior reports, it has become clear that TGF-β maintains both tumor suppressing and promoting effects, depending on the model.⁶² Some have suggested that TGF-β is initially tumor suppressive, but becomes promoting as the stage of disease progresses.⁶¹

Transforming growth factor β and its protein family members have been associated with aggressive and invasive tumors, as well as associated with the activation of the stellate cell that leads to pancreatic desmoplasia. Family member proteins are multifunctional and are known to be associated with a variety of different functions, including immune cell regulation, migration, and cell proliferation.⁶³ The TGF-β family of ligands includes TGF-β1, TGF-β2, and TGF-β3. Transforming growth factor β–mediated signaling propagates through 1 of 3 TGF-β receptors (TGF-βRI to TGF-βRIII) and Smad signaling proteins. In many cases, TGF-β binds to a heterodimer of TGF-βRI and TGF-βRII.

Smad4/DPC4 has been reportedly inactivated in more than 50% of pancreatic tumors. The apparent discrepancy between the correlation of TGF-β and poor outcomes in patients and the finding that Smad4/DPC4 is inactivated in the cancer cells of the majority of PDAC patients suggests that SMAD4-independent pathways play an important role in the progression of some tumors.⁶⁴ Wild-type Smad4/DPC4 appears to be associated with decreased invasion and improved outcomes in patients.^65,66 Some of the Smad-independent TGF-β signaling pathways that have been reported include Ras, PI3K, RhoA, and MAP3K1.^67–70 Some of the early findings on the role of TGF-β in cancer came from tetracycline-inducible transgenic MMTV–TGF-β mouse models. In their report, Muraoka-Cook and colleagues⁷¹ found a 10-fold enhanced rate of lung metastases following oncogenic transformation by MMTV-PyVT expression and TGF-β induction. Despite the worries that the pleiotropic nature of TGF-β may preclude therapeutic development and intervention, at least in cancer, some initial studies report that might not be the case.^72–74 Of note, TGF-β haploinsufficiency itself can inhibit excessive fibrosis and progression of precancerous lesions in preclinical models of pancreatic cancer.⁷⁵ We recently assessed the effects of the TGF-βRI inhibitor, LY2157299, on tumor stroma using a genetically engineered mouse model for pancreatic cancer. As shown in Figure 1, tumors in mice treated with the combination of LY2157299 and gemcitabine showed a marked reduction in overall stromal content; hematoxylin-eosin staining, top panel) and collagen I deposition (bottom panel) when compared with vehicle-treated mice or gemcitabine-only–treated mice. Our results suggest that TGF-βR inhibition may be an effective means of ameliorating the effects of desmoplasia in pancreatic cancer, which may in turn improve the efficacy of current treatment regimens. With these findings, TGF-β antagonists are currently under investigation as therapeutic agents in a variety of indications (e.g., galunisertib [LY2157299], clinicaltrials.gov, NCT01373164), including hepatocellular carcinoma, pancreatic cancer, or myelodysplastic syndromes.

Epithelial-To-Mesenchymal Transition

Additional hypotheses to address problems associated with desmoplasia include targeting EMT that is induced by TGF-β expression and secretion by cancer-associated fibroblasts (CAFs). It is generally understood that mesenchymal cells are generally more resistant to drug treatment than epithelial cells. Indeed, erlotinib resistance in head and neck squamous cell carcinoma is correlated to increased Zeb-1 expression and decreased E-cadherin expression. Increased Zeb-1 and decreased E-cadherin are classic markers of EMT, and these changes follow TGF-β ligand binding and signal transduction.^76,77 Targeting EMT may prove to be a viable approach as multiple studies have shown improved antitumor efficacy, such as with the mucin-reactive PAM4 antibody or with the secreted clusterin–reactive antibody AB-16B5 (clinicaltrials. gov, NCT02412462).^78–80

Secreted Protein, Acidic, And Rich In Cysteine

SPARC (secreted protein, acidic, and rich in cysteine) is an albumin-binding protein that is overexpressed in multiple types of cancer.^81–83 SPARC expression noted in the stroma in pancreatic cancer but not in tumor cells has been associated with poor survival.^84–86 It has been noted that nanoparticle albumin (nab) paclitaxel had shown clinical antitumor activity in several cancer types that overexpressed SPARC.^87,88 In a 66-patient pilot trial combining nab-paclitaxel with gemcitabine, an overall response rate (complete plus partial response) of 45% was noted with a median survival of 11.2 months and a 1-year survival of 48%.¹⁶ In that trial, the expression of SPARC in the stroma but not in the tumor was correlated with improved survival.

In a multicenter, multiple-country, 861-patient randomized trial of nab-paclitaxel + gemcitabine versus gemcitabine as a single agent, the 2-drug regimen gave a response rate of 23% versus 7% (with the single-agent gemcitabine), median survival of 8.5 months versus 6.7 months (hazard ratio, 0.72, P < 0.001), and 35% survival at 1 year versus 22%, 9% versus 4% at 2 years, and 4% long term (>3 years) for that combination versus 0% on the gemcitabine-alone control arm.^7,89 Hidalgo and colleagues⁹⁰ utilized a different antibody treatment to determine if the presence of SPARC by immunohistochemistry predicted for survival with the nab-paclitaxel + gemcitabine regimen. Despite confirming the results of the pilot trial,¹⁶ Hidalgo and colleagues⁹⁰ could not find an association between SPARC levels and efficacy in patients with metastatic pancreatic cancer enrolled in this definitive 861-patient study (although only 256 patients had enough specimen to have stromal SPARC be evaluated).

In summary, although finding SPARC in pancreatic cancer led to evaluation of nab-paclitaxel because it was believed that SPARC in the tumor stroma would cause accumulation of nab-paclitaxel in the tumor, a very large study did not confirm that to be the case. We are grateful, however, that the nab-paclitaxel + gemcitabine has now become one of the standard of care regimens for patients with advanced pancreatic cancer.

Of special interest is a follow-up trial by Alvarez and colleagues.⁹¹ In that trial using endoscopic ultrasound elastography on patients with advanced pancreatic cancer treated with nab-paclitaxel + gemcitabine, they noted a decrease in stiffness of the tumor and on biopsies noted decreases in density of CAFs as well as decreased and disorganized collagen. Their conclusion was that nab-paclitaxel + gemcitabine regimen caused a remarkable alteration of cancer stroma that resulted in “tumor softening.”⁹¹