Clinical Trial of Oral Nelfinavir before and during Radiation Therapy for Advanced Rectal Cancer

Pelvic radiotherapy (RT) has an important role in the treatment of patients with rectal adenocarcinoma. Short course RT, 25 Gy delivered in 5 daily fractions in 1 week followed by surgery within 5 to 7 days, can halve the risk of local recurrence in patients with operable rectal cancer (1, 2). Long-course preoperative chemoradiotherapy (LCCRT), typically 45 to 50.4 Gy in 25 to 28 daily fractions over 5 to 6 weeks in combination with 5-fluorouracil or capecitabine as a radiosensitizer, is generally offered to patients with locally advanced tumors. Tumor regression has been shown to correlate with improved outcomes for patients (3–5).

The optimal first treatment for patients with a symptomatic primary rectal cancer and distant metastases at presentation is a matter of debate. Systemic therapy is not effective in all patients; although it may achieve response after 6 to 8 weeks of therapy, it does not provide rapid symptom relief for all patients (6). Planning and delivery of LCCRT may delay delivery of full-dose systemic therapy and may therefore compromise surgical treatment of metastatic disease (e.g., liver surgery for operable metastases). A strategy of short-course RT followed 2 weeks later by full-dose systemic combination chemotherapy can be used to prevent this delay. Short-course RT can safely precede full-dose systemic therapy (e.g., capecitabine and oxaliplatin and bevacizumab), resulting in pathological complete response (pCR) rates above 25% and radical resection and/or radiofrequency ablation of all metastatic disease in the majority of patients (7).

One factor that increases cellular resistance to RT is overexpression of activated oncogenes, such as the epidermal growth factor receptor (EGFR; ref. 8), RAS (9), or loss of the tumor suppressor gene PTEN (10). These mutations share molecular signaling via the PI3K–Akt pathway. We have previously shown that inhibition of this pathway augments response to RT in vitro and in vivo in cells with constitutive activation of this pathway, an effect not seen in cells with a nonactivated pathway (11–14). This pathway is frequently altered in humans with colorectal cancer (15). Because the PI3K signaling pathway can be constitutively activated in tumor cells, yet not in host cells, an inhibitor of this pathway might be expected to improve the therapeutic index through selective tumor radiosensitization (16).

Nelfinavir is an HIV protease inhibitor (HPI) that has been shown to inhibit Akt at standard clinical doses and to cause radiosensitization in vivo (17). In addition to intrinsic radiosensitization, we have shown previously that nelfinavir caused sustained improvements in tumor perfusion and reduction in hypoxia in a mouse xenograft model (18). Although some clinical studies have investigated nelfinavir in combination with chemoradiotherapy (see Table 1), there are no published data on the addition of nelfinavir to RT without concomitant chemotherapy. Nor are there data on whether the changes in perfusion observed in preclinical studies with nelfinavir are replicated in human subjects with cancer. Dynamic contrast-enhanced MRI (DCE-MRI) and perfusion CT (p-CT) have previously been used to detect changes in tumor perfusion induced by anti-angiogenic drugs (19, 20) and chemoradiotherapy for rectal cancer (21–25).

A barrier to the advancement of radiosensitizers is uncertainty regarding the optimal primary endpoint for clinical trials. Endpoints traditionally used, such as pCR rate, radiological response or disease-free survival, have a number of limitations, including variability of definitions (26). The development of new tissue biomarkers of response is highly desirable for the evaluation of novel radiosensitizers. We have developed a quantitative assessment of tumor cell density (TCD), which is a predictor of survival in patients with colorectal cancer (27). We are currently exploring this technique to compare different preoperative RT schedules (28).

The objective of the SONATINA (Study Of Nelfinavir Addition to Radiotherapy Treatment In Neo-Adjuvant Rectal Cancer) clinical trial was to investigate the safety of nelfinavir administered before and during RT in patients with rectal adenocarcinoma. We also explored the feasibility of incorporating biomarkers of RT that could be used in efficacy studies and the ability of p-CT and DCE-MRI to detect changes in tumor perfusion during therapy.

SONATINA was a nonrandomized, open-label clinical trial (EudraCT number: 2010-020621-40) to establish the safety of nelfinavir with hypofractionated pelvic RT. The primary outcome was measured by the occurrence of any grade 3 or higher toxicities [Common Terminology Criteria for Adverse Events (CTCAE), version 4.0] within 28 days of the last fraction of RT. Because the primary outcome was the safety of this novel combinatorial therapy, there was no control group. Secondary outcomes included radiological response of primary tumor at 8 weeks after RT, feasibility of measuring a tissue biomarker (TCD) in pretreatment biopsies and biopsies taken 7 days after RT, and feasibility of using dynamic imaging to evaluate tumor perfusion.

Ethical approval was obtained from National Research Ethics Service Committee South Central (reference 10/H0604/61). Key inclusion criteria were patients with histologically proven adenocarcinoma of the rectum, radiological evidence of M1 disease, suitability for short-course RT as primary treatment (determined by the Colorectal Tumor Board), ECOG performance status 0 to 2, and age ≥ 18 years. Exclusion criteria included previous pelvic RT, recent severe cardiac disease, or operable primary tumor (in opinion of the Tumor Board).

Patients received 7 days of oral Nelfinavir (1,250 mg b.i.d.) before RT and a further 7 days of nelfinavir during RT. This dose of nelfinavir has been shown to consistently reduce levels of Akt phosphorylation in peripheral blood mononuclear cells in patients with cancer (29). Compliance logs were used to check that all doses were taken as prescribed. The total dose of RT was 25 Gy, delivered in 5 Gy fractions on 5 days during a 7-day period as a single-phase treatment prescribed to the International Commission on Radiation Units (ICRU) Reference Point. The dose constraints were set such that at least 99% of the planning target volume (PTV) should receive 95% of the prescription dose. The PTV maximum was no more than 107% of the prescribed dose to the ICRU reference point. For all patients, 3 to 7 photon beams (6 or 15 MV) were used, with the entire plan displayed in physical dose. Conformal RT plans were reviewed by a RT quality assurance panel (independent clinician, radiographer, physicist) prior to delivery of the first fraction. Verification imaging by cone beam CT to localize the treatment volume was required prior to every fraction for the first three fractions. In order to treat metastases, patients were permitted to commence systemic chemotherapy 14 days after the completion of RT.

Patients underwent MRI of the pelvis at baseline and 8 weeks after completion of RT for assessment of tumor regression grade (mrTRG) according to a recognized scoring system (30). As previously published (30), patients with mrTRG score of 1 to 3 on MRI scan were classified as having “good mrTRG score” and patients with mrTRG score of 4 or 5 were classified as having “poor mrTRG score.” Anonymized scans were assessed by two independent radiologists; agreement was evaluated by weighted Kappa statistic. In cases of discrepancy, scans were assessed by a third independent radiologist and consensus was derived.

In order to explore changes in tumor perfusion induced by protocol therapy, DCE-MRI and p-CT scans of the rectum were incorporated at three time points: before commencement of nelfinavir, the day before commencement of RT (i.e., day 7 of nelfinavir), and on the last day of the treatment (before the RT fraction was delivered). Mean p-CT parameters [blood flow, blood volume, and mean transit time (MTT)] and median DCE-MRI scan parameters (K^trans, K^ep, and V^e) were measured in the tumor volume of interest, and percentage change in these values was presented graphically.

In diagnostic biopsies and biopsies performed 7 days after completion of RT, TCD was measured in digitally scanned hematoxylin and eosin–stained slides using an automated scanning system (Aperio XT, Aperio Technologies) at 200× magnification (27, 28). In cases where there was variation in TCD across the specimen, we used the area of tumor with highest TCD, as we have previously reported and correlated with clinical outcomes (27). Immunohistochemistry was carried out on pretreatment rectal biopsy specimens using the Leica Bond-Max automated immunostainer (Leica Microsystems) on 5-μm sections cut from formalin-fixed paraffin-embedded tissue. As an indicator of baseline characteristics, pretreatment biopsy sections were stained for the following biomarkers: CAIX, HIF1α, Phospho-PRAS40 (see Supplementary Information).

The Wilcoxon signed-rank test was used to determine pairwise differences for nonparametric data, and the paired Student t test was used to determine pairwise differences for parametric data.

From April 2011 to August 2013, 19 patients were screened and 10 patients were recruited (Fig. 1; Table 2). All patients completed RT as per protocol. Compliance logs revealed that 1 patient missed one dose of nelfinavir and another patient missed two doses of nelfinavir.

There were no grade 4 toxicities. Two patients stopped taking nelfinavir early because of toxicity: one on day 13 of treatment because of an allergic rash (grade 3, probably related), the other on day 4 due to vomiting (grade 3, possibly related but the patient had preexisting partial gastric outlet obstruction). Additionally, 5 patients had grade 3 toxicities within 28 days of RT (Table 3). One patient was admitted to hospital with grade 3 diarrhea 23 days after the completion of RT and nelfinavir, which was 7 days after the commencement of oxaliplatin and 5FU chemotherapy. This event was considered to be related to chemotherapy and possibly related to RT, but unrelated to nelfinavir. Another patient developed grade 3 diarrhea 4 days after the completion of nelfinavir and RT; this event was considered to be causally related to protocol therapy. Another patient had grade 3 perianal pain due to hemorrhoids, probably related to RT.

With regard to laboratory values, one patient developed grade 3 lymphopenia on the last day of protocol therapy. This persisted on a blood test 1 month following completion of therapy. The total white cell count was normal, and the patient had no evidence of active infection. A number of grade 1 or 2 abnormalities in liver function tests were observed within 3 months of therapy, likely to be related to liver metastases or chemotherapy (Supplementary Table S1). One patient had hyponatremia (grade 3), which preceded protocol therapy, and worsened transiently during an episode of diarrhea after RT. Because a known side effect of nelfinavir is diabetes mellitus, fasting glucose was checked during treatment and follow-up. Three patients had grade 1 or 2 hyperglycemia after 7 days of nelfinavir; blood glucose was normal on subsequent testing 28 days after completion of therapy.

Using a recognized scoring system (30), interobserver agreement between two independent radiologists was good, with weighted kappa score of 0.79. Of 9 patients who completed MRI scans of the pelvis 8 weeks after completion of nelfinavir and RT to assess mrTRG response of the primary tumor, 5 patients exhibited “good” tumor regression according the definitions of the scoring system (ref. 30; Table 4; Supplementary Fig. S1). It should be noted that, as discussed in the Introduction, a major benefit of the treatment strategy adopted in this clinical trial was that patients were permitted to commence full-dose systemic chemotherapy to treat metastatic disease as early as 14 days from the last fraction of RT, as documented in Table 4. 

All 10 patients in the study successfully completed p-CT scans at 3 time points (Supplementary Fig. S2). The pCT scans for 1 patient (patient 7) were excluded from analysis for technical reasons. Nine patients underwent DCE-MRI scanning at all 3 time points. One patient (patient 1) did not undergo the second DCE-MRI scan because of vertigo. A further 3 scans were excluded from analysis because of inadequate contrast enhancement or contrast extravasation.

Analyzing the percentage change in perfusion parameters between the pretreatment scans (scan 1) and the scan on the seventh day of nelfinavir (scan 2), the median blood flow was 37.3 at scan 1, and 43.9 at scan 2 (nonsignificant by the Wilcoxon signed-rank test). There were also no statistically significant changes in blood volume or MTT demonstrated between scans 1 and 2 (nonsignificant by the Wilcoxon signed-rank test).

Between the p-CT on the seventh day of nelfinavir (scan 2) and the scan at the end of RT (scan 3), an increase in blood flow in association with a decrease in MTT was observed in 8 of 9 evaluable patients (Fig. 2A). A significant median 30% increase in blood flow (P = 0.01, Wilcoxon signed-rank test) and a 29% median decrease in MTT was observed (P = 0.01, Wilcoxon signed-rank test) on p-CT from scan 2 to scan 3 (Supplementary Table S2).

Between the DCE-MRI on the seventh day of nelfinavir (scan 2) and the scan at the end of RT (scan 3), an increase in median K^trans was demonstrated in all 7 evaluable patients (Fig. 2B; Supplementary Table S3). Between scans 2 and 3, there was a 42% (0.08/minute) mean increase in median K^trans and a 13% (0.07) mean increase in median V^e (P = 0.03 and 0.02, respectively, Student t test).

TCD was evaluable in all of the pretreatment rectal biopsy specimens and in 9 out of 10 post-radiotherapy biopsy specimens (Fig. 2C). The median TCD decreased from 24 (interquartile range, 13–45) at baseline to 9 (interquartile range, 3–16) on post-treatment biopsies. One of the post-treatment biopsies contained adenoma cells but no malignant cells, which was attributed to sampling error; this sample was not included in analyses.

The sample size was not adequate to study potential relationships between somatic or immunohistochemical analyses (Supplementary Fig. S3) at baseline and radiological response 8 weeks from the end of RT, but these data are presented in Table 4 and Supplementary Tables S4 to S6 because they may assist in the design of future studies of this treatment combination. Of note, 7 of 10 tumors had KRAS mutation.

Nelfinavir has been shown to inhibit Akt at standard clinical doses and to cause radiosensitization in vivo (17). This early-phase trial was designed to study the safety of nelfinavir with hypofractionated pelvic RT and to develop both tissue and imaging biomarkers of the potential efficacy of this combinatorial therapy for use in future studies. We have demonstrated that the combination of nelfinavir and hypofractionated pelvic RT is well tolerated in patients with advanced rectal cancer.

Although the sample size in this study was not sufficient to make any definite conclusions about the response rate, the proportion of good mrTRG in the study presented here compares favorably with LCCRT for locally advanced rectal cancer. In one large UK study, the rate of good mrTRG for LARC was 50% overall (30) and for ≥T3c tumors only 33%. This compares with 56% in the study presented here, in which 60% patients had T4 tumors and 70% had a KRAS mutation. It should be noted that 4 of the patients with good mrTRG score had 3 to 6 weeks of chemotherapy between the end of RT and MRI assessment. Although systemic therapy may have contributed to the clinical response rates observed, the ability to administer full-dose systemic therapy soon after RT appears to be a promising treatment strategy with regard to clinical response rates. The efficacy of hypofractionated RT followed by systemic chemotherapy in comparison with standard chemoradiation is currently being tested in the international, multicenter, randomized trial RAPIDO (NCT01558921; ref. 31).

Importantly, the SONATINA study is the first clinical trial to assess the safety of nelfinavir and RT without the confounding effect of concurrent chemotherapy (see Table 1). A previous study of nelfinavir and long-course chemoradiotherapy with capecitabine resulted in unacceptable levels of grade 3 hepatotoxicity (32), which may have been attributable to a drug interaction between chemotherapy and nelfinavir. Similarly, in a study of concurrent nelfinavir, temozolomide, and RT for patients with glioma, 3 patients experienced dose-limiting grade 3 transaminase elevation (33). In our study, we observed 3 grade 3 toxicities that were considered to be possibly or probably related to nelfinavir: diarrhea, drug rash, and lymphopenia. Of these, only the drug rash was a dose-limiting toxicity. Consistent with the published toxicities of hypofractionated pelvic RT without nelfinavir (34–36), our conclusion is that the addition of nelfinavir to hypofractionated pelvic RT is well tolerated. Importantly, hepatotoxicity was not observed in our study (see Table 1; Supplementary Information). It should be noted that 7 of 10 patients treated in the clinical trial reported here had low rectal tumors (Table 2). We propose that future studies including patients with mid and high rectal tumors should carefully document toxicities to ensure the safety of treating larger volumes of small intestine with RT.

In addition to intrinsic radiosensitization, we have shown previously that nelfinavir caused sustained improvements in tumor perfusion and reduction in hypoxia in a mouse xenograft model after 5 to 14 days of treatment (18). We therefore evaluated two imaging biomarkers to measure potential changes in perfusion during nelfinavir therapy in patients with cancer: p-CT and DCE-MRI. Although no changes were observed from 7 days of the trial drug, our study showed a 30% increase in mean blood flow using p-CT and a 42% mean increase in median k^trans using DCE-MRI scans during RT and nelfinavir. The intrasubject coefficient of variation for blood flow in colorectal tumors has been reported to be in the range of 14% to 23% (37, 38), and studies suggest that the coefficient of variation for k^trans measurements in tumors using DCE-MRI is of the order of 20% (39, 40). In our study, the consistency between the findings of the two imaging modalities adds substantial support to the observation of increased tumor perfusion. Although k^trans can be affected by permeability, our findings from p-CT as well as DCE-MRI suggest increased blood flow from the combination of nelfinavir plus RT.

Because there was no control group (i.e., no nelfinavir) in this early-phase trial designed to show the safety of protocol therapy, it is not possible to differentiate the effect of RT on blood flow from the effect of nelfinavir plus RT in the data from our imaging biomarkers. Previous studies of LCCRT have demonstrated increases in tumor perfusion parameters during the initial weeks of RT (22, 41) followed by subsequent decreases in tumor perfusion after completion of therapy (21, 24, 42–44). Our findings are consistent with previously reported increases in median k^trans between baseline and the fifth fraction of hypofractionated RT for locally advanced rectal cancer (23). In order to ascertain whether the significant changes we have observed are due to RT or due to the combination of nelfinavir with RT, we propose that phase II studies of the efficacy of nelfinavir-RT versus RT alone should incorporate imaging biomarkers of blood flow.

At present, tissue biomarkers for the selection of patients for a treatment strategy including a novel radiosensitizing drug do not exist. Visual estimation of the tumor:stroma ratio has been shown to be prognostic for patients with localized colon cancer (45), but this has not been studied in patients with metastatic rectal cancer scheduled to receive RT. We sought to develop a reproducible, quantitative tissue biomarker of potential radiosensitization for use in future clinical trials. We have previously assessed TCD in pretreatment biopsy specimens and resected tumors (27, 28), and in the study presented here we assessed the feasibility of measuring TCD in both pre-RT and post-RT biopsy samples obtained at endoscopy.

Our results are in favor of the hypothesis that the addition of nelfinavir to hypofractionated RT may result in additional tumor cell kill compared with RT alone. Compared with our previous study of 45 rectal cancer patients who received 25 Gy in 5 fractions of RT to the pelvis followed by surgery 7 days after the end of radiotherapy (28), whose TCD values ranged from 14 to 46, the range of post-treatment TCDs in this study was 1 to 21. Based on these findings, we conclude that TCD can be measured in biopsies taken pre- and post-RT. Although TCD could be developed further as a biomarker of radiosensitizing drugs for use in prospective clinical trials, there are limitations in assessing TCD from biopsies due to differences in sampling techniques. Larger, correlative studies with imaging such as mrTRG are warranted.

This study has shown that the combination of nelfinavir and hypofractionated RT for locally advanced rectal cancer is well tolerated, and that this novel treatment strategy can be followed by combination chemotherapy as early as 14 days after RT to treat metastatic disease. Consistent with previous studies of RT, nelfinavir plus hypofractionated RT significantly increased mean blood flow to tumor compared with baseline values. The tissue biomarker TCD can be measured on biopsies taken before and after RT; it is a candidate biomarker for systematic development for assessing potential radiosensitizing drugs prior to phase II evaluation.

P. Quirke reports receiving speakers bureau honoraria from Amgen and Roche, and reports receiving commercial research support from Leica and Roche. R. Muschel is a consultant/advisory board member for Midatech, and reports receiving commercial research support from Galaxy Biotech. No potential conflicts of interest were disclosed by the other authors.

Conception and design: E.J. Hill, L. Boyle, E.M. Anderson, S. Dutton, S.B. Love, J.A. Schnabel, R. Muschel, W.G. McKenna, R.A. Sharma

Development of methodology: E.J. Hill, M. Enescu, N. West, E.M. Anderson, G. Brown, S.B. Love, P. Quirke, W.G. McKenna, R.A. Sharma

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): E.J. Hill, N. West, T.P. MacGregor, K.-Y. Chu, L. Boyle, C. Blesing, L.-M. Wang, E.M. Anderson, G. Brown, P. Quirke, R.A. Sharma

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): E.J. Hill, C. Roberts, J.M. Franklin, M. Enescu, T.P. MacGregor, G. Brown, S. Dutton, S.B. Love, J.A. Schnabel, P. Quirke, M. Partridge, R.A. Sharma

Writing, review, and/or revision of the manuscript: E.J. Hill, C. Roberts, J.M. Franklin, M. Enescu, N. West, T.P. MacGregor, K.-Y. Chu, L. Boyle, C. Blesing, L.-M. Wang, S. Mukherjee, E.M. Anderson, G. Brown, S. Dutton, S.B. Love, J.A. Schnabel, P. Quirke, W.G. McKenna, M. Partridge, R.A. Sharma

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): C. Roberts, L. Boyle, L.-M. Wang, W.G. McKenna

Study supervision: E.M. Anderson, W.G. McKenna, R.A. Sharma

Other (contributed patients to the study): S. Mukherjee

The authors thank the patients who participated in this study, the University of Oxford as sponsor, present and former staff at Oncology Clinical Trials Office and Centre for Statistics in Medicine (both part of the UKCRC registered Oxford Clinical Trials Research Unit (OCTRU), members of the Independent Early Phase Trials Oversight Committee for support and guidance, Andrew Slater, Margaret Betts, Emma Tinkler-Hundal and the Rectal Subgroup of the NCRI Colorectal Clinical Study Group and NCRI CTRad Working Group for advice during protocol development.

This study was supported by Oxfordshire Health Services Research Committee, CRUK-ESPRC Oxford Cancer Imaging Centre, NIHR Oxford Biomedical Research Centre, Oxford ECMC, Pathological Society of Great Britain and Ireland, Academy of Medical Sciences, Higher Education Funding Council for England, Yorkshire Cancer Research.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.