The American Cancer Society statistics for 2014 estimate 136,830 new cases of colorectal cancer (CRC) in the United States [1] . CRC is the third leading cause of cancer death-related mortality with over 50,000 deaths in the United States and fourth worldwide with around 700,000 deaths annually [2] . The importance of CRC among various cancers is also reflected by the number of new treatments introduced in the last three decades, beginning with 5-fluorouracil in the early 1980s and followed by its derivative capecitabine 15 years later. Today, standard first-line treatments include fluorouracil with leucovorin and irinotecan or oxaliplatin, alone or combined with bevacizumab. Cetuximab, the immunoglobulin G1 monoclonal antibody against the epidermal growth factor receptor (EGFR), is effective in combination with irinotecan in patients with metastatic colorectal cancer or as a single agent in patients with metastatic colorectal cancer that progresses even when irinotecan is used. Phase 1 and 2 studies have also shown cetuximab to have activity when added to irinotecan-based therapy or oxaliplatin- based therapy as first-line treatment. Other monoclonal antibodies, bevacizumab, panitumumab, and ziv- aflibercept have been added to treatment regimens over the last ten or so years, and more recently the multikinase inhibitor regorafenib [3] . As the result of these treatment advances, resectable CRC has become a treatable disease with surgery in the early stages, or with surgery followed chemotherapy. Regimens such as FOLFOX, FOLFIRI, XELOX, and XELIRI are used as first- and second-line treatment for the management of advanced stage IV disease. Unfortunately, few options are available after failure of second-line therapy.

The private practice at the Burzynski Clinic (BC) in Houston, Texas, has focused its interest on treatment of advanced cases of cancer that failed standard therapy. Results of treatment of such patients in both clinical trials and private practice have recently been published and indicate that response rates and overall survival in cases of malignant brain tumor, advanced pancreatic cancer, and mesothelioma may be improved [4] -[12] . Our research attention has been focused on the anti-cancer activity of antineoplastons (ANP) and sodium phenylbutyrate (PB) as treatment adjuncts [13] -[19] .

PB, a histone deacetylase (HDAC) inhibitor is the salt of an aromatic fatty acid used in the treatment of urea cycle disorders [17] . It is currently being explored in combination with cytotoxics and other novel drugs. Derived from its HDAC activity, PB is being investigated for use as a potential differentiation-inducing agent in malignant glioma, acute promyelocytic leukemia and many other disorders [19] . ANP is a group of anti-cancer agents that are peptides, amino acid derivatives, and carboxylic acids originally isolated from blood and urine of healthy people [13] -[16] . Some constituents of ANP, namely phenylacetylglutaminate (PG) and phenylacetate (PN) are also metabolites of PB, which in addition to receiving approval for urea cycle disorders is being investigated for adjunctive use in glioma and acute promyelocytic leukemia [10] [16] -[19] . A study of PG and PN on the neoplastic genome of glioblastoma multiforme (GBM) reveals that they affect over 100 genes [20] . Modeling from the GBM genome study and molecular profiling was used to formulate treatment plans for CRC patients.

The evaluation of patients described in this article provides treatment data of 15 patients with advanced CRC who failed at least two lines of standard treatment.

Fifteen subjects were diagnosed with advanced CRC at outside institutions. Radiology and pathology studies were performed by institutions not associated with BC. This group included all consecutive evaluable patients treated at the BC between June 3, 2004 and December 13, 2011.

Laboratory tests were performed by both the laboratory at the BC and outside facilities. Tests included standard blood and urine analysis and the determination of genomic markers. Molecular profiling based on tumor tissue analysis was performed by Foundation Medicine, Cambridge, MA and Caris Life Science, Phoenix, AZ. All patients were provided with treatment details and were required to sign an informed consent document prior to receiving therapy. A treatment plan based on molecular profiling included PB in combination with targeted and chemotherapeutic agents. Patients were treated on an out-patient basis, and after the initial two to four weeks of treatment at the BC, they continued therapy under the care of a local oncologist. Before initiating treatment, a baseline computerized tomography (CT) scan, with and without contrast, and in some patients, a positron emission tomography (PET) scan was performed. The products of two of the largest perpendicular diameters (LPD) of the largest measurable lesions were calculated and totaled providing a baseline evaluation for each study subject. The baseline provided a reference for determining response outcomes to the treatment. Additional pretreatment measurements included Karnofsky Performance Status (KPS), vital signs, clinical disease status, demographics, medical history and current medications, physical examination, and EKG. Evaluation of toxicity was performed according to the Common Terminology Criteria for Adverse Events, version 3 (CTC- AEv.3). Potential responses to treatment include complete response (CR), partial response (PR), stable disease (SD), and progressive disease (PD). CR required the disappearance of all lesions confirmed at the end of four weeks, PR required 50% or higher decrease of the LPD of measurable lesions confirmed at four weeks, PD was determined when there were new lesions or when there was an increase over 25% in the existing lesions, and SD was classified as the status between PR and PD. The duration for each response was measured from the date the criteria of the outcome were first met until the date that PD was first documented. In the case of SD, the duration was measured from the time the therapy commenced.

This paper describes a potential strategy for a more successful treatment of colorectal cancer after failure of second line therapy. Over 25 years ago, a model for colorectal carcinogenesis was introduced [24] . Epigenetic

changes and mutations in oncogenes and tumor suppressor genes were instrumental in occurrences of gradual abnormalities leading from small adenomatous polyps to invasive adenocarcinoma. Multiple genetic alterations included activation of KRAS, interference in WNT signaling and mutation of TP53 [24] . EGFR signaling played a major role in the development of therapy for CRS [25] . A major breakthrough was the introduction of monoclonal anti-EGFR antibodies, cetuximab and panitumumab [26] . Despite initial success, the resistance toward anti-EGFR antibodies frequently developed due to mutation and amplification of EGFR or genes in the downstream pathways [27] -[30] . A critical member of the downstream RAS pathway is Kerstin-RAS oncogene (KRAS) [31] . The elucidation of the role of KRAS mutations in the efficacy of anti-EGFR therapies became a major area of oncology research. Patients with mutated KRAS seldom responded to cetuximab and panitumumab [32] [33] . Recent data, however, indicate that recurrence in patients with wild-type KRAS treated with anti-EGFR antibodies depends on overgrowth of undetected KRAS mutant cell population, and it was concluded that the majority (93%) of CRC tumors have KRAS mutations [31] [34] . Differences in the median OS in wild-type and KRAS mutated groups of patients are shown in Table 4 [21] [22] .

Abbreviations: OS-median overall survival.

In addition to monoclonal antibodies that block the extracellular domain of EGFR, the tyrosine kinase inhibitors (TKI), erlotinib and gefitinib that target the intracellular domain of the receptor were assessed in clinical studies [3] . Unfortunately, there were no objective responses and no increased OS in phase II studies with either erlotinib or gefitinib with FOLFIRI [35] [36] . In first-line therapy, however, the addition of erlotinib with bevacizumab following induction with chemotherapy and bevacizumab resulted in statistically significant increase in PFS regardless of KRAS mutation [37] .

Due to frequent mutations, KRAS offered a good theoretical therapeutic target. Tipifarnib, an inhibitor of farnesylation of RAS protein failed in CRC clinical trials as a result of conversion from farnesylation to geranylation in cancer cells [38] . BRAF inhibition fared much better. Oral multikinase inhibitor, regorafenib, inhibits BRAF and VEGFR, PDFRβ, FGFR, KIT and RET [39] . Phase III trial with regorafenib confirmed a median OS of 6.4 months [23] . Drug resistance, however, occurs early which is supported by loss of function of PTEN and activation of AKT, and anti-apoptosis action through suppression of BAD [40] . Additional mechanisms of resistance to BRAF inhibitors include secondary KRAS mutations and activation of MEK/ERK [41] [42] . MEK inhibitor selumetinib produced encouraging results, but caused resistance through up-regulation through WNT signaling [43] [44] .

The importance of m-TOR in the regulation of growth of cancer cells suggests m-TOR inhibitors are good candidates for clinical trials, a promise that was not fulfilled in studies with sirolimus [45] [46] . An additional concern was the activation of PI3K signaling through a negative feedback loop [47] . The new generation of both m-TOR and PI3K inhibitors demonstrated better results in combination with erlotinib in PTEN negative tumors [48] .

Epigenetic mechanisms leading to CRC were studied for a number of years [49] . The agents that promote demethylation of promoters of the tumor suppressor genes and HDAC inhibitors were proposed for treatment of CRC [50] . PB, ANP and vorinostat can be considered in this respect [19] [20] .

The failure of combination chemotherapy to eradicate advanced CRC is attributed to persistence of cancerous stem cells (CSC) [51] [52] . Mutations that are crucial in carcinogenesis of CRC, including inactivation of tumor suppressor gene adenomatous polyposis coli (APC) and activation of KRAS, give competitive advantage to stem cells and convert them into the neoplastic process [52] . Genes of developmental pathways participate in the process of formation and maintenance of CSC [53] .

Wingless-related integration site (WNT), bone morphogenic protein (BMP), Notch and hedgehog (HH) pathways are critical in this process [53] . Inhibition of WNT signaling through Cox-2 inhibitors celecoxib, indomethacin and aspirin did not produce a significant effect [54] [55] . On the other hand, the activation of non-canon- ical WNT signaling by HDAC inhibitors inhibited APC mutated CRC cells [56] . This indicates the possible use of vorinostat and PB which in a different study have shown cytotoxic effect against CSC [57] . m-TOR inhibitor everolimus was also found to inhibit WNT signaling [58] .

The data from several research centers suggest a general activation of Notch-1 signaling in CRC [59] -[61] . Despite initial promising results, the phase II trials with γ-secretase inhibitor RO4929097 had little effect [62] [63] .

The evidence that inhibition of signal transduction in BMP and HH pathways may find clinical application was not yet presented [53] .

Additional approaches to control CRC may include interference in cell cycle, cancer cell metabolism, angiogenesis, and inhibition of apoptosis. These mechanisms were explored in the treatment of recurrent GBM using PB in combination with targeted agents [10] .

Treatment of advanced CRC after failure of a second-line of standard-of-care therapy creates a challenge for oncologists. The current view on KRAS carcinogenesis recognizes a high background level of KRAS mutation in patients who tested positive for the wild-type gene [31] . Anti-EGFR monotherapies lead to outgrowth of resistant KRAS mutants and recurrence [31] .

Based on available research data and our limited experience, we propose the following strategy for the combination of targeted and chemotherapeutic agents (Figure 4 and Figure 5). It is suggested to simultaneously interrupt EGFR (VEGFR)-KRAS-ERK and PI3K-AKT pathways. Patients with wild-type KRAS not previously treated with anti-EGFR antibodies have a reasonable chance to respond to cetuximab or panitumumab in combination with other drugs. Alternatively, based on studies by others, erlotinib can be used synergistically with bevacizumab and chemotherapy regardless of KRAS status [37] . For patients who failed FOLFIRI, capecitabine, and/or oxaliplatin can be reconsidered. In our evaluation, multikinase inhibitors, pazopanib and sorafenib, were used since regorafenib was not yet available. In a new strategy, regorafenib can be used instead. New MEK and

PI3K inhibitors may be considered once they become available. The effect against CSC can be accomplished through a combination of HDAC inhibitors, vorinostat and PB, in conjunction with m-TOR inhibitors, everolimus/sirolimus [57] [58] . Additional targeted agents can be considered based on molecular profiling.

Treatment with PB in combination with targeted agents and chemotherapy appears to provide another option for improved outcomes in patients who failed two lines of standard chemotherapy for advanced CRC. Such combination treatment typically requires a dosage reduction, since toxicities of the drugs can overlap. We understand that the findings with PB in combination with targeted agents and/or chemotherapy are preliminary and we propose validation of these data using a well-designed phase I/II trial in advanced recurrent CRC. This principle for CRC patients who fail prior treatment may also have validity using ANP, which possess overlapping ingredients with metabolites of PB as they have shown promise in the treatment of various brain tumors, including GBM [4] -[10] . ANP can offer the advantage of higher anticancer activity, since they are available in intravenous dosage form. Randomized, controlled clinical trial in advanced CRC patients comparing intra-arterial chemotherapy for the treatment of liver metastases with ANP plus intra-arterial chemotherapy have been successfully completed and have shown 5-year-survival statistics twice as good as the control (intra-arterial chemotherapy). The report is now in press.

Despite rapid progress in the treatment of CRC, the established standard-of-care for patients who recurred after the second-line chemotherapy provides poor results, with progress being very modest. The use of targeted agents as a single treatment or in combination with chemotherapy has not provided substantial survival benefit. The results reported here are based on a small series of patients who were consecutively admitted for the treatment at BC during the last few years. This is a retrospective evaluation that shows an increase of median OS and tolerable toxicity compared to other available treatments. The choice of targeted agents was limited when this evaluation began. Furthermore, molecular profiling was in its early stages, providing limited data that were helpful for the design of treatment plans. The group included only evaluable patients, which is typical for a retrospective analysis. The authors realize that results are preliminary and the sample size is small. They should be validated in a larger population by well-designed phase I/II clinical trials with PB or ANP in combination with targeted or chemotherapeutic agents. Caution should be exercised when combining these agents, since no clinical trials have been conducted yet with such combinations. We also propose that future clinical trials include molecular profiling to help select the subgroups of cases of CRC and coordinate genomic changes with responses.

The authors express their appreciation to the additional physicians involved in the care of the patients: Drs. Zanhua Yi, Alejandro Marquis, Robert Weaver, Sheryl Acelar, Lourdes De Leon, and Mohammad Khan. Preparation of the manuscript was provided by Carolyn Powers, Jennifer Pineda, Elizabeth Cleveland, Maria Corzo, and Adam Golunski.

Written informed consent was obtained from patients for publication of this case report and accompanying images.

All authors are employed by Burzynski Clinic. Dr. Stanislaw R. Burzynski and Dr. Gregory S. Burzynski are shareholders and directors, and Dr. Tomasz J. Janicki is the Vice-President of Burzynski Research Institute, Inc. Dr. Stanislaw R. Burzynski is President of Burzynski Research Institute, Inc., Dr. Gregory S. Burzynski is Vice-President of Burzynski Clinic and Dr. Sheldon Brookman is Director of Pharmaceutical Development of Burzynski Clinic.