Session 4: Shaping radiotherapy for rectal cancer: should this be personalized?

Preoperative radiotherapy continues to be widely used in patients with operable rectal cancer. However, the indications and goals for such treatment are evolving. Professor Marijnen reviews the historic and current evidence base for the use of preoperative neoadjuvant radiotherapy and the future challenges in tailoring the therapy according to the patients' needs and tumour stage.

Impact of multimodal therapy in locally recurrent rectal cancer.

BACKGROUND: The practice of salvaging recurrent rectal cancer has evolved. The aim of this study was to define the evolving salvage potential over time among patients with locally recurrent disease, and to identify durable determinants of long-term success. METHODS: The study included consecutive patients with recurrent rectal cancer undergoing multimodal salvage with curative intent between 1988 and 2012. Predictors of long-term survival were defined by Cox regression analysis and compared over time. Re-recurrence and subsequent treatments were evaluated. RESULTS: After multidisciplinary evaluation of 229 patients, salvage therapy with curative intent included preoperative chemotherapy and/or radiotherapy (73·4 per cent; with 41·3 per cent undergoing repeat pelvic irradiation), surgical salvage resection with or without intraoperative irradiation (36·2 per cent), followed by postoperative adjuvant chemotherapy (38·0 per cent). Multivisceral resection was undertaken in 47·2 per cent and bone resection in 29·7 per cent. The R0 resection rate was 80·3 per cent. After a median follow-up of 56·5 months, the 5-year overall survival rate was 50 per cent in 2005-2012, markedly increased from 32 per cent in 1988-1996 (P = 0·044). Long-term success was associated with R0 resection (P = 0·017) and lack of secondary failure (P = 0·003). Some 125 patients (54·6 per cent) developed further recurrence at a median of 19·4 months after salvage surgery. Repeat operative rescue was feasible in 21 of 48 patients with local re-recurrence alone and in 17 of 77 with distant re-recurrence, with a median survival of 19·8 months after further recurrence. CONCLUSION: The long-term salvage potential for recurrent rectal cancer improved significantly over time, with the introduction of an individualized treatment algorithm of multimodal treatments and surgical salvage. Durable predictors of long-term success were R0 resection at salvage operation, avoidance of secondary failure, and feasibility of repeat rescue after re-recurrence.

p53 tumor suppressor gene status and the degree of genomic instability in sporadic colorectal cancers.

BACKGROUND: Genomic instability reflects the propensity and the susceptibility of the genome to acquire multiple alterations and, in turn, is believed to be a driving force behind multistep carcinogenesis. Although the molecular basis of genomic instability in sporadic colorectal cancers remains largely a mystery, mutation of the p53 tumor suppressor gene (also known as TP53) has been proposed to play an integral role in this process. However, a dilemma exists in that p53 mutation appears to be a late event in the progression of sporadic colorectal tumors, whereas genomic instability, serving as a facilitator of tumor progression, is envisioned as occurring early in this process. PURPOSE: We evaluated the relationship between p53 mutation and the major form of genomic instability in sporadic colorectal tumors, namely, that involving DNA breakage, which leads to chromosomal translocations, insertions, deletions, and gene amplification. METHODS: Fifty-eight sporadic colorectal tumors that had been previously evaluated for genomic instability were analyzed for p53 mutations. These tumors were from consecutively diagnosed patients. Genomic instability was quantified by use of inter-simple sequence repeat polymerase chain reaction analysis that employed (CA)8RG and (CA)8RY primers (R = purine [A or G]; Y = pyrimidine [C or T]); a genomic instability index (a measure of the number of alterations in tumor DNA in comparison with normal DNA, expressed as a percent) was calculated for each tumor. Mutation of the p53 gene in exons 5-9 was determined by use of single-strand conformational polymorphism-polymerase chain reaction analysis and DNA sequencing. Chi-squared analysis was used to determine the statistical significance of differences between groups of tumors. Reported P values are two-sided. RESULTS: p53 mutations were identified in 29 (50%) of the 58 tumors. The median genomic instability index value was 3.3%. Nineteen (65.5%) of the 29 tumors with p53 mutations had genomic instability indices that were less than the median value (range, 0%-2.6%); the remaining 10 (34.5%) tumors had genomic instability indices that were greater than the median (range, 3.9%-13.0%). Eleven (37.9%) of the 29 tumors with wild-type p53 genes had genomic instability indices that were less than the median value (range, 0%-2.6%), whereas the remaining 18 tumors had genomic instability indices above the median (range, 3.9%-11.7%). There was a statistically significant association between a lesser degree of genomic instability and the presence of p53 mutations (P = .032). CONCLUSIONS AND IMPLICATIONS: Tumors with no or minimal evidence of genomic instability are more likely to harbor p53 mutations than tumors with evidence of substantial genomic instability. p53 mutations play an important role in the development of cancers but do not appear to initiate or promote genomic instability in sporadic colorectal tumors.

Somatic frameshift mutations in DNA mismatch repair and proapoptosis genes in hereditary nonpolyposis colorectal cancer.

An exacerbated genomic instability at simple repeated sequences characterizes cancer of the microsatellite mutator phenotype (MMP). The majority of hereditary nonpolyposis colon cancers (HNPCCs) and about 15% of nonselected ("sporadic") gastrointestinal tumors belong to the MMP pathway of tumorigenesis. Colorectal MMP+ and MMP- tumors exhibit fundamental differences in genotype and phenotype. We have shown previously that "sporadic" MMP+ colon cancers exhibit a paradoxical low incidence of somatic mutations in the p53 tumor suppressor gene and the c-K-ras proto-oncogene. On the other hand, gastrointestinal MMP+ cancers frequently harbor frameshift mutations in genes containing mononucleotide repeats. These include the cell growth regulator gene TGFbetaRII and the proapoptotic gene BAX. We have also recently shown the frequent presence of frameshift mutations in (A)8 and (C)8 tracts within the hMSH3 and hMSH6 DNA mismatch repair genes in sporadic colon cancer of the MMP. Here, we describe the nearly identical incidence of somatic frameshift mutations in these genes in a panel of 27 HNPCC MMP+ cancers: 52% in hMSH3 and BAX and 33% in hMSH6. In contrast, no mutations in any of these genes were found in 10 MMP- cancers of HNPCC patients. These results show that the multistep model for the unfolding of the MMP also applies to HNPCC and further illustrate the importance of the escape from apoptosis in the MMP pathway for gastrointestinal cancer. They also underscore the differences in genotype between tumors with and without enhanced microsatellite instability and the similarities in genotype between tumors of the MMP regardless of their hereditary or sporadic nature.

An anoxia inducible endonuclease and enhanced DNA breakage as contributors to genomic instability in cancer.

Fischer rat embryo fibroblasts subjected to temporary anoxia followed by an aerobic recovery period show genomic instability in the form of highly elevated CAD gene amplification rates. As revealed by flow cytometric analysis this is associated with DNA breakage in vivo, followed by repair during the recovery period. Such genomic instability parallels expression of a M(r) 29,000/31,000 endonuclease; this enzyme requires no added divalent metal ion and has a pH optimum of about 6.5. The same endonuclease was found to be expressed within healing wounds and in four of ten human colorectal cancers but was not seen in eight normal colorectal tissue samples. Our results indicate that DNA breakage resulting from endogenous endonuclease activity can have a substantial effect in modulating genomic instability.

A National Cancer Institute Workshop on Microsatellite Instability for cancer detection and familial predisposition: development of international criteria for the determination of microsatellite instability in colorectal cancer.

In December 1997, the National Cancer Institute sponsored "The International Workshop on Microsatellite Instability and RER Phenotypes in Cancer Detection and Familial Predisposition," to review and unify the field. The following recommendations were endorsed at the workshop. (a) The form of genomic instability associated with defective DNA mismatch repair in tumors is to be called microsatellite instability (MSI). (b) A panel of five microsatellites has been validated and is recommended as a reference panel for future research in the field. Tumors may be characterized on the basis of: high-frequency MSI (MSI-H), if two or more of the five markers show instability (i.e., have insertion/deletion mutations), and low-frequency MSI (MSI-L), if only one of the five markers shows instability. The distinction between microsatellite stable (MSS) and low frequency MSI (MSI-L) can only be accomplished if a greater number of markers is utilized. (c) A unique clinical and pathological phenotype is identified for the MSI-H tumors, which comprise approximately 15% of colorectal cancers, whereas MSI-L and MSS tumors appear to be phenotypically similar. MSI-H colorectal tumors are found predominantly in the proximal colon, have unique histopathological features, and are associated with a less aggressive clinical course than are stage-matched MSI-L or MSS tumors. Preclinical models suggest the possibility that these tumors may be resistant to the cytotoxicity induced by certain chemotherapeutic agents. The implications for MSI-L are not yet clear. (d) MSI can be measured in fresh or fixed tumor specimens equally well; microdissection of pathological specimens is recommended to enrich for neoplastic tissue; and normal tissue is required to document the presence of MSI. (e) The "Bethesda guidelines," which were developed in 1996 to assist in the selection of tumors for microsatellite analysis, are endorsed. (f) The spectrum of microsatellite alterations in noncolonic tumors was reviewed, and it was concluded that the above recommendations apply only to colorectal neoplasms. (g) A research agenda was recommended.

Genomic DNA-based hMSH2 and hMLH1 mutation screening in 32 Eastern United States hereditary nonpolyposis colorectal cancer pedigrees.

Hereditary nonpolyposis colorectal cancer (HNPCC) is an autosomal dominant cancer syndrome characterized by early age of onset colorectal cancer (mean 45 years) as well as endometrial, urinary tract, and upper gastrointestinal adenocarcinomas. The HNPCC phenotype has been shown to segregate with germline mutations in the human homologues of the DNA mismatch repair genes MSH2, MLH1, PMS1, and PMS2. However, the majority of published DNA mismatch repair gene mutation surveys associated with HNPCC kindreds report multiple levels of preselection, including 2p and 3p chromosomal linkage analysis and the evaluation of microsatellite instability of proband colorectal cancers prior to mutation analysis. For this reason, the concise contribution of each of the known DNA mismatch repair genes to the HNPCC phenotype remains unknown. We report the results of a genomic DNA-based analysis of hMSH2 and hMLH1 germline mutations in 32 unrelated Eastern United States HNPCC kindreds. These families were selected for study on the basis of phenotype only. We identified three hMSH2 and six hMLH1 mutations in eight families, for a positive mutation rate of 25%. Two mutations were identified in one of the families. Four of the mutations detected have not been reported in the literature previously. One of the mutation-positive families is African American; the others were of European-American ancestry. These results provide a clarification of the contribution of hMSH2 and hMLH1 to the HNPCC phenotype and suggest that in the majority of Eastern United States HNPCC kindreds selected by phenotype alone, the molecular genetic basis for the disease remains unknown.

Intrachromosomal genomic instability in human sporadic colorectal cancer measured by genome-wide allelotyping and inter-(simple sequence repeat) PCR.

We have used genome-wide allelotyping with 348 polymorphic autosomal markers spaced, on average, 10 cM apart to quantitate the extent of intrachromosomal instability in 59 human sporadic colorectal carcinomas. We have compared instability measured by this method with that measured by inter-(simple sequence repeat) PCR and microsatellite instability assays. Instability quantitated by fractional allelic loss rates was found to be independent of that detected by microsatellite instability analyses but was weakly associated with that measured by inter-(simple sequence repeat) PCR. A set of seven loci were identified that were most strongly associated with elevated rates of fractional allelic loss and/or inter-(simple sequence repeat) PCR instability; these seven loci were on chromosomes 3, 8, 11, 13, 14, 18, and 20. A lesser association was seen with two loci flanking p53 on chromosome 17. Coordinate loss patterns for these loci suggest that at least two separate sets of cooperating loci exist for intrachromosomal genomic instability in human colorectal cancer.

High prevalence of Inoue-Melnick virus antibodies in patients with colorectal carcinoma.

The high prevalence of serum neutralizing antibodies against Inoue-Melnick virus (IMV) among American patients with colorectal carcinoma has been confirmed. Sera from 26 patients with colorectal carcinoma along with the identical number of age- and sex-matched patients with non-colorectal neoplasia and normal healthy controls were collected in the Buffalo area. All of the colorectal carcinoma group possessed antibodies against IMV (100%), while antibody positivity for non-colorectal neoplasia and for normal controls were 34.6% and 38.5%, respectively. Geometric mean titers of antibodies to IMV type 1 and type 2 for colorectal carcinoma were 266 and 338, respectively, whereas the mean titers in the other two control groups were less than 10.3. These differences between colorectal carcinoma and the two controls were highly significant (P less than 0.001). The majority of patients with colorectal carcinoma had antibodies to both IMV types 1 and 2.

The reliability of immunohistochemistry as a prescreening method for the diagnosis of hereditary nonpolyposis colorectal cancer (HNPCC)--results of an international collaborative study.

Hereditary nonpolyposis colorectal cancer syndrome (HNPCC) is an autosomal dominant condition accounting for 2-5% of all colorectal carcinomas as well as a small subset of endometrial, upper urinary tract and other gastrointestinal cancers. An assay to detect the underlying defect in HNPCC, inactivation of a DNA mismatch repair enzyme, would be useful in identifying HNPCC probands. Monoclonal antibodies against hMLH1 and hMSH2, two DNA mismatch repair proteins which account for most HNPCC cancers, are commercially available. This study sought to investigate the potential utility of these antibodies in determining the expression status of these proteins in paraffin-embedded formalin-fixed tissue and to identify key technical protocol components associated with successful staining. A set of 20 colorectal carcinoma cases of known hMLH1 and hMSH2 mutation and expression status underwent immunoperoxidase staining at multiple institutions, each of which used their own technical protocol. Staining for hMSH2 was successful in most laboratories while staining for hMLH1 proved problematic in multiple labs. However, a significant minority of laboratories demonstrated excellent results including high discriminatory power with both monoclonal antibodies. These laboratories appropriately identified hMLH1 or hMSH2 inactivation with high sensitivity and specificity. The key protocol point associated with successful staining was an antigen retrieval step involving heat treatment and either EDTA or citrate buffer. This study demonstrates the potential utility of immunohistochemistry in detecting HNPCC probands and identifies key technical components for successful staining.

Primary adenocarcinoma arising in an ileostomy.

Two cases are added to the world literature of patients in whom an adenocarcinoma developed at the ileostomy site after total proctocolectomy for ulcerative colitis. Fourteen additional cases have been reported in the world literature; of these, 12 cases have been in patients with ulcerative colitis, and four cases have been in patients with familial adenomatous polyposis. Adenocarcinoma of an ileostomy is not common. However, in the analysis of the reported cases, patients with long-standing ileostomies appear to be at a greater risk. With an aging ileostomy population, an increase in the number of cases may be seen. Three hypotheses are discussed as potential causative pathways to this entity. Continued analysis of these cases may yield information on the pathophysiology involved.

Photodynamic therapy in the management of neoplasms of the perianal skin.

Perianal extramammary Paget's disease, Bowen's disease, and squamous cell carcinoma are three entities that are very rarely reported. Overall, it is generally accepted that wide surgical excision is adequate treatment except under certain circumstances. Consideration has to be given to invasiveness, metastatic potential, multicentricity, and tendency to recur. We describe four patients who, following unsuccessful multiple attempts at achieving microscopically clear margins with surgical excision, were treated at Roswell Park Cancer Institute with photodynamic therapy with a surface illumination method. With a minimum follow-up of 6 months (more than 1 year in three patients), there have been no recurrences to date.

99mTc-IMMU-4 monoclonal antibody scan in colorectal cancer. A prospective study.

A blinded prospective study of 34 patients with colorectal adenocarcinoma using the Fab' fragment of the anticarcinoembryonic antigen monoclonal antibody type IMMU-4 labeled with technetium 99m was conducted to compare, on a lesion-by-lesion basis, the findings of radioimmunoscintigraphy, preoperative computed tomography, and exploratory celiotomy. Of 115 lesions detected at surgery, 113 were adenocarcinoma. Radioimmunoscintigraphy detected 59 lesions and computed tomography detected 62; both studies combined detected 72. Twenty-nine (54%) lesions missed by radioimmunoscintigraphy and 24 (45%) missed by computed tomography were 1 cm or smaller. When both studies were combined, the sensitivities were 90%, 24%, and 42%, and the specificities were 52%, 86%, and 61% for hepatic, extrahepatic intra-abdominal, and pelvic lesions, respectively. In 10 patients, additional information obtained with the radioimmunoscintigram could have altered the treatment of these patients. In this study, radioimmunodetection scan was complementary to computed tomographic scan in the examination of patients with colorectal carcinoma.

Absorbable mesh sling prevents radiation-induced bowel injury during "sandwich" chemoradiation for rectal cancer.

HYPOTHESIS: Absorbable mesh slings can prevent radiation-induced bowel injury when adjuvant pelvic radiotherapy is given in the early postoperative period. We hypothesized that the mesh sling technique is similarly effective during "sandwich" sequence adjuvant chemoradiation. DESIGN: Retrospective review. SETTING: Tertiary care comprehensive cancer center. PATIENTS: Nonrandomized series of 19 consecutive patients who underwent abdominoperineal resection and received postoperative sandwich sequence chemoradiation at Roswell Park Cancer Institute, Buffalo, NY, between January 1994 and September 1999. INTERVENTIONS: Twelve patients had an absorbable mesh sling placed at the completion of abdominoperineal resection. Seven patients did not have an absorbable mesh sling placed. MAIN OUTCOME MEASURES: Radiotherapy dose and gastrointestinal toxic effects. RESULTS: All 12 patients in the "mesh" group were able to receive full-dose radiotherapy with tumor bed boost (total dose, 54 Gy, 11 patients; 59.4 Gy, 1 patient). Only 3 of 7 patients in the "no mesh" group were able to receive a tumor bed boost (total dose, 46.8 Gy, 1 patient; 50.4 Gy, 3 patients; 54 Gy, 3 patients). Acute gastrointestinal toxic effects were minimal in the mesh group (grade 1, 10 patients; grade 2, 2 patients) compared with the no mesh group (grade 2, 6 patients; grade 3, 1 patients). None of the patients in the mesh group have shown evidence of late gastrointestinal toxic effects. One patient in the no mesh group required surgery for complications of chronic radiation enteritis. CONCLUSIONS: The protective effects of an absorbable mesh sling extend beyond the life expectancy of the mesh itself. Sandwich sequence chemoradiation should not preclude the use of the mesh sling technique.

A phase I and pharmacokinetic study of oral uracil, ftorafur, and leucovorin in patients with advanced cancer.

A phase I and pharmacokinetics study of oral uracil, ftorafur, and leucovorin was performed in patients with advanced cancer. Uracil plus ftorafur (UFT) was given in a 4:1 molar ratio in three divided doses for 28 consecutive days. Patient cohorts were treated at 200, 250, 300, and 350 mg/m2 of UFT daily. For all patients, 150 mg of leucovorin was given daily in three oral doses. A 1-week rest period followed each 28-day treatment course. Gastrointestinal toxicity, characterized by diarrhea, nausea, and vomiting, was dose-limiting at 350 mg/m2 UFT in patients who had received prior chemotherapy. Mild fatigue and transient hyperbilirubinemia were also common. In previously untreated patients, UFT at 350 mg/m2 was well-tolerated, suggesting this as an acceptable phase II dose in this schedule with leucovorin. Two of eight previously untreated patients with advanced colorectal cancer had partial responses with UFT (350 mg/m2) plus leucovorin. Pharmacokinetic parameters [ftorafur, uracil, 5-fluorouracil (5-FU), 5-methyltetrahydrofolate] showed wide interpatient variations. Plasma levels of 5-FU (Cmax 1.4 +/- 1.9 microM) were comparable to those achieved with protracted venous infusions, and folate levels (Cmax 6.1 +/- 3.6 microM) were sufficient for biochemical modulation. Ongoing study will determine if this convenient oral regimen will compare favorably in terms of efficacy, toxicity, and cost with intravenous fluoropyrimidine programs.

Phase I and pharmacokinetic evaluation of floxuridine/leucovorin given on the Roswell Park weekly regimen.

A phase I and pharmacokinetics study was carried out of floxuridine (FdUrd) modulated by leucovorin (LV) given on the Roswell Park regimen (LV given at 500 mg/m2 by 2-h infusion and FdUrd given by i.v. push at 1 h after the start of LV infusion, treatment being given weekly x 6). The dose-limiting toxicity was diarrhea; the MTD and recommended dose for phase II studies was 1,650 mg/m2 per week of FdUrd. The dose-response curve was steep, with 3/3 patients treated at a dose of 1,750 mg/m2 developing grade IV diarrhea. With this schedule there was no significant mucositis. Pharmacokinetic parameters showed very wide interpatient variability. Plasma decay was biphasic with a t1/2 beta of approximately 2 h. Plasma clearance was high (> 200 1 h-1). No correlation between pharmacokinetic parameters and toxicity could be identified.

Anal canal and perianal epidermoid cancers.

Squamous cell carcinoma of the anal canal serves as a paradigm for the successful application of multimodality treatment of solid tumors. Since 1974, multimodality treatment with combined radiation and chemotherapy has become the standard. The compelling advantage of sphincter preservation and the substantial survival benefit compared with surgery alone prompted investigators to adopt chemoradiation treatment. Several questions regarding the optimal radiation dose and chemotherapy in initial as well as salvage therapy remain, and only recently have results of several prospective randomized studies become available to address some of these unresolved issues. We reviewed the clinical aspects and historical treatment results of anal canal and perianal epidermoid cancers in light of the results of these modern trials. Current management strategies are redefined, and future directions of clinical studies are outlined.

The management of infectious and noninfectious anorectal complications in patients with leukemia.

BACKGROUND: Infectious and noninfectious anorectal complications may occur in patients undergoing therapy for leukemia. Controversy surrounds the treatment of this problem in immunocompromised patients. STUDY DESIGN: A retrospective review of the medical records of 83 patients with acute or chronic leukemia in whom anorectal disease developed during inpatient therapy for leukemia was performed to determine the initial signs and symptoms, treatment, and outcomes. RESULTS: During a 12-year period, 92 patients with anorectal complications were treated. This series included 25 patients with perirectal abscesses, 22 patients with anal fissures, 18 patients with symptomatic external hemorrhoids, 12 patients with perianal ulcerations, 12 patients with symptomatic internal hemorrhoids, and three patients with fistulas in ano. Overall, 79 (86 percent) of the 92 anorectal complications resolved in 68 of the 83 patients. Increasing periods of neutropenia did not adversely affect the resolution of anorectal disease. Thirteen patients (16 percent) required surgical intervention, most commonly secondary to a perirectal abscess. Incision and drainage was necessary in ten (40 percent) of 25 patients with perirectal abscess, which included five patients with fluctuation and five patients in whom infection failed to respond to medical therapy. CONCLUSIONS: Noninfectious anorectal complications in patients with leukemia respond to nonoperative intervention and rarely progress to a life-threatening infection. Nonoperative intervention in the form of systemic antibiotics and sitz baths is successful in the treatment of infectious anorectal complications. Incision and drainage should be performed when fluctuation is present and in patients whose complications fail to respond to medical therapy.

Endoscopic laser treatment for palliation of colorectal adenocarcinoma.

Fifty-three patients were treated for palliation with endoscopic neodymium yttrium aluminum garnet (Nd:YAG) laser for colorectal carcinoma at Roswell Park Cancer Institute. There were 25 females and 28 males. The mean age of the patients was 76 years. Thirty-eight tumours were primary and 15 recurrent. The level of the lesions from the anal verge ranged from 2 to 50 cm, with a mean of 10 cm. Eighty-four percent of the lesions were rectal carcinomas within 15 cm of the anal verge. Lesion length ranged from 1.5 to 12 cm with a mean of 5 cm. The number of laser treatments ranged from 1 to 17 with a mean of 3. The duration of treatments ranged from 25 to 90 min with a mean of 35 min. The mean number of joules per treatment was 5093. Eight patients (15%) developed complications. There were no mortalities. The success rate for treating bleeding, the most frequent presenting symptom, was 93%. The overall success rate of patients improved by laser treatment was 79%. Patients who were improved by therapy had a median survival of 18 months versus those not improved who had a median survival of 3 months (P = 0.01). The use of the Nd:YAG laser for palliative treatment of patients with colorectal carcinoma is safe, effective and associated with no mortality.

Altered immunoglobulin A (IgA-1) levels in patients with advanced colorectal adenocarcinoma presenting with normal preoperative levels of carcinoembryonic antigen (CEA).

We investigated the value of serum IgA-1 as a complementary tumour marker to carcinoembryonic antigen (CEA) in the monitoring of the postoperative follow-up of 19 patients with advanced colorectal carcinoma presenting with normal levels of CEA. Mean follow-up period was 14 months (range 2-72 months). Mean number of serum specimens was 5 (range 2-9). IgA-1 assay employed rabbit antihuman IgA for binding IgA-1 from patient's serum and peanut agglutinin to detect the IgA-1 O-glycosidic structure. Ten patients developed distant metastases while nine patients had locoregional recurrence. All 19 patients had generally persistent elevation of serum IgA-1 throughout the follow-up period while serum CEA levels were subsequently elevated in only 10 patients. IgA-1 predicted recurrence with an average lead time of 8 months (range 1-19 months). On the other hand, mean lag time for CEA was 12 months (range 6-50 months). The data indicate the clinical utility of serum IgA-1 as a potential complementary tumour marker to CEA in the monitoring of the postoperative course of this particular subset of colorectal cancer patients.