- Email: [email protected]
What lies beyond the view of the sigmoidoscope?
March 2001
SELECTED SUMMARIES
the data and issues involved in that choice, although not addressed in our paper, were discussed by Dr. Koretz. We agree with many of Dr. Koretz’s points. First, what we and others may label as “advanced lesions” are really “surrogate markers” whose clinical significance (i.e., natural history to develop into CRC) may be speculated on but is not known. We agree that the “marginal cost-effectiveness” of colonoscopy compared with sigmoidoscopy and other screening modalities is a critically important consideration before choosing which procedure, if any, is preferred for CRC screening. As part of that assessment, safety data need to be fully considered. And we agree that the implementation of mass colonoscopy screening would need to be carefully organized because it would involve an enormous redirection of resources, direct dollar cost, and “opportunity cost” of physicians and other services which would, at least temporarily, be diverted from other health care activities. Dr. Koretz correctly points out that there are no clinical trials (which provide the strongest evidence of efficacy) to support either colonoscopic or sigmoidoscopic screening. However, we do not agree that the next step is to do a “large RCT to assess the true efficacy of avoiding having polyps left behind.” Such a trial would be lengthy, cumbersome, expensive, and is unlikely to add much, if anything, to the currently available “indirect” evidence that strongly supports the efficacy of colonoscopic screening. Such evidence comes from the RCTs of fecal occult blood test screening and the case control studies of sigmoidoscopy screening. Further, 2 clinical trials of screening sigmoidoscopy are now under way. We believe that, although additional clinical data about safety and cost-effectiveness need to be generated and analyzed, it is unnecessary to do a randomized controlled trial. Last, despite uncertainties about the details of CRC screening, we suggest that the glass should be seen as three quarters full rather than one quarter empty. A variety of screening modalities seems to be effective and cost-effective, and it seems prudent to start to implement any of them (N Engl J Med 2000;343:1641–1643). THOMAS F. IMPERIALE, M.D. DAVID F. RANSOHOFF, M.D.
WHAT LIES BEYOND THE VIEW OF THE SIGMOIDOSCOPE? Lieberman DA, Weiss DG, Bond JH, Ahnen DJ, Garewal H, Chejfec G, for the Veterans Affairs Cooperative Study Group 380 (Veterans Affairs medical centers in Portland, Oregon, Perry Point, Maryland, Minneapolis, Minnesota, Denver, Colorado, Tucson, Arizona, and Hines, Illinois). Use of colonoscopy to screen asymptomatic adults for colorectal cancer. N Engl J Med 2000;343:162–168. Imperiale TF, Wagner DR, Lin CY, Larkin GN, Rogge JD, Ransohoff DF (Department of Medicine, Indiana University Medical Center, Roudebush Veterans Affairs Medical Center, Indianapolis Gastroenterology Research Foundation, Eli Lilly, Indianapolis, Indiana; and Department of Medicine, the University of North Carolina at Chapel Hill, Chapel Hill, North Carolina). Risk of advanced proximal neoplasms in asymptomatic adults according to the distal colorectal findings. N Engl J Med 2000;343:169 –174.
1051
The two methods most commonly used to screen for colorectal cancer and adenomatous polyps are fecal occult blood testing (FOBT) and flexible sigmoidoscopy. Well-designed randomized controlled trials (Lancet 1996;348:1472–1477, 1996;348:1467–1471, N Engl J Med 1993;328:1365–1371) and case-control studies (J Natl Cancer Inst 1992;84:1572– 1575, N Engl J Med 1992;326:653– 657, Ann Intern Med 1995;123:904 –910) provide convincing evidence of the efficacy of both of these screening methods. However, both methods are less sensitive than colonoscopy for detection of colorectal neoplasms. In particular, flexible sigmoidoscopy– based screening strategies, in which only patients with neoplasms in the distal colon undergo colonoscopy, are prone to missing isolated neoplasms in the proximal colon. As a result, there is growing interest in colonoscopy as a primary screening test for average-risk patients. Two recently completed studies have addressed the potential miss rate of screening flexible sigmoidoscopy for neoplasms in the proximal colon. Lieberman et al. reported the results of a cohort study of screening colonoscopy performed at 13 Veterans Affairs medical centers (N Engl J Med 2000;343: 162–168). They enrolled 3196 patients without symptoms of lower intestinal disease, a history of colitis, colonic polyps, or colorectal cancer, and without a colonic examination (sigmoidoscopy, barium enema, or colonoscopy) within the preceding 10 years. All patients underwent colonoscopy. If the colonoscopy was not completed to the cecum, a repeat colonoscopy was attempted within 6 months of the first colonoscopy. All polyps identified at colonoscopy were removed for pathologic evaluation. Adenomatous neoplasms measuring at least 10 mm in diameter and containing villous histology, high-grade dysplasia, or invasive cancer were categorized as advanced neoplasms. The distal colon was classified as the rectum, sigmoid colon, and descending colon in the primary analysis, and as the rectum and sigmoid colon in a secondary analysis. From 4659 eligible patients, 3196 patients (96.8% men) enrolled in the study. The mean age was 62.9 years. Colonoscopy was completed to the cecum in 97.7% of the patients. Adenomatous neoplasms and advanced neoplasms were identified in 1171 patients (37.5%) and 329 patients (10.5%), respectively. Among the 128 patients with advanced proximal lesions, 62 (48.4%) had at least one adenoma distal to the splenic flexure. Using the alternative definition of the distal colon, 64 of 169 patients (37.9%) with proximal advanced neoplasms had synchronous distal adenomas. The presence of either small or large adenomas in the distal colon was positively associated with presence of advanced proximal lesions (odds ratio [OR], 2.6; 95% confidence interval [CI ], 1.7– 4.1 for small lesions; OR, 3.4, 95% CI, 1.8 – 6.5 for large lesions). Imperiale et al. reported nearly identical results in a similarly designed study (N Engl J Med 2000;343:169 –174). They performed screening colonoscopy on employees of Eli Lilly aged 40 or older with no lower gastrointestinal symptoms, no history of colonic neoplasms, and no history of inflammatory bowel disease. The distal colon was defined as the rectum, sigmoid colon, and descending colon. Advanced
1052
SELECTED SUMMARIES
neoplasms were defined as adenomas with villous features, high-grade dysplasia, or invasive cancer. Analyses were limited to subjects 50 years or older at the time of screening. A total of 1994 patients (58.9% men) were included in the analysis. The mean age was 59.8 years. Colonoscopy was completed to the cecum in 97% of patients. Advanced proximal neoplasms were identified in 50 patients (2.5%), of whom the most advanced distal finding was an advanced neoplasm in 7 (14%), a tubular adenoma in 12 (24%), a hyperplastic polyp in 8 (16%), and no polyp in 23 (46%). Compared with persons without distal polyps, advanced proximal lesions were more common in persons with distal hyperplastic polyps (OR, 2.6; 95% CI, 1.1–5.9), tubular adenomas (OR, 4.0; 95% CI, 1.9 – 8.3), and advanced neoplasms (OR, 6.7; 95% CI, 3.2– 16.6). Analyses using an alternative outcome of a proximal neoplasm at least 10 mm in diameter yielded nearly identical results. Comment. Both studies are thoughtful and well designed. Both used sound inclusion criteria and clinically meaningful outcome measures, had low rates of loss to follow-up (i.e., incomplete colonoscopy), and used appropriate statistical analytic techniques. Furthermore, the studies had nearly identical results, with more than half of all patients with advanced proximal neoplasms having no adenomas in the distal colon. Thus a logical interpretation of these studies is that, although the presence of distal colonic polyps is associated with an increased prevalence of advanced proximal colonic neoplasms, more than half of all advanced neoplasms in the proximal colon would be missed by a strategy of screening with sigmoidoscopy alone. Before we accept this conclusion, it is important to consider the potential limitations of these studies. One potential limitation is the use of the colonoscopist’s assessment of polyp location to determine the proportion of polyps that would be identified at screening flexible sigmoidoscopy. It is often difficult to state with certainty the exact anatomic location from which a polyp is removed. Given that the endoscopists were aware of the study hypothesis, this raises a small potential for bias. If the endoscopists consistently overestimated the proximal position of polyps (e.g., estimating that polyps in the descending colon were in the transverse colon), this could lead to an overestimate of the flexible sigmoidoscopy miss rate for proximal polyps. However, the consistent findings of these 2 studies (N Engl J Med 2000;343:162–168 and 2000;343:169 –174), as well as the similar findings of other studies (Gastrointest Endosc 1996;44:109 – 111 and 1999;49:727–730) suggest that this is unlikely to be the case. Another limitation of the studies is that the anatomic depth of examination by flexible sigmoidoscopy can be quite variable (Endoscopy 1999;31:227–231). Assuming that all sigmoidoscopies reach the splenic flexure may underestimate the proportion of advanced proximal neoplasms that would be missed. The secondary analysis in which the distal colon was defined as the rectum and sigmoid colon emphasizes this point (N Engl J Med 2000;343:162–168). Screening programs using FOBT and sigmoidoscopy rely on colonoscopic polypectomy for efficacy. As such, it seems inevitable that screening colonoscopy programs should produce greater reductions in cancer-related mortality than the other strategies. However, these studies were not designed to prove that screening colonoscopy is superior to flexible sigmoidoscopy every 5 years and/or annual FOBT for reducing colorectal cancer–related mortality. Because these were cross-sectional studies, we have no information on the natural
GASTROENTEROLOGY Vol. 120, No. 4
history of the lesions identified. Similarly, we do not know whether new polyps would have formed in the distal colon of persons with advanced proximal neoplasms before these became invasive cancers. These studies suggest that screening sigmoidoscopy to the splenic flexure with full colonoscopy for any patient with an adenoma of any size in the distal colon will identify approximately 80% of prevalent advanced colonic neoplasms (N Engl J Med 2000;343:162–168). Thus, the expected gain of a strategy of screening colonoscopy compared with flexible sigmoidoscopy would potentially be limited to the 20% of cases not identified with sigmoidoscopy. Combining FOBT with flexible sigmoidoscopy would be expected to identify a proportion of the advanced neoplasms missed by the screening sigmoidoscopy. In addition, some advanced lesions may be identified before they progress to invasive cancer if polyps are identified on repeat sigmoidoscopy or at colonoscopy completed for the evaluation of new symptoms. Thus, the net gain of screening colonoscopy compared with flexible sigmoidoscopy every 5 years plus annual FOBT may be relatively small, and formal comparison of these strategies in a randomized controlled trial may be impractical. In addition, the cost of obtaining this small gain in life expectancy may be quite large. Two recent cost-effective analyses both suggest that screening colonoscopy every 10 years will yield greater life expectancy at a reasonable cost per year of life gained than screening with sigmoidoscopy alone or fecal occult blood testing alone (Ann Intern Med 2000;133:573–584, JAMA 2000;284:1954 –1961). However, the analysis of Frazier et al. suggests that a strategy of annual FOBT in conjunction with sigmoidoscopy every 5 years was the most cost-effective. Furthermore, in their analysis, the combination of annual FOBT and flexible sigmoidoscopy every 5 years yields greater life expectancy than screening colonoscopy every 10 years, assuming 60% compliance with screening recommendations (JAMA 2000;284:1954 –1961). Importantly, the estimate of the most costeffective strategy was sensitive to both the compliance rate and the cost of endoscopic procedures. Because these are both moving targets, it is not possible to state with certainty whether colonoscopy or the combination of flexible sigmoidoscopy and FOBT is the most effective or the most cost-effective strategy. Certainly, compliance with annual FOBT and sigmoidoscopy every 5 years is often lower than 60% (Mor Mortal Wkly Rep CDC Surveill Summ 1999;48:116 – 121). In populations in which compliance with repeated testing is low, the long-term protection of ever having undergone colonoscopy may be much greater than ever having undergone flexible sigmoidoscopy or FOBT. Several other issues relevant to a strategy of screening colonoscopy require further attention. Firstly, how frequently should screening colonoscopy be performed? Similarly, how long can we extend the period between surveillance colonoscopies for persons with a history of adenomatous polyps? Finally, who will perform these colonoscopies? In 1993, Ransohoff and Lang estimated that a strategy of screening flexible sigmoidoscopy every 5 years would require 250,000 physician-days per year to screen all eligible persons in the United States and additional person-hours to complete the follow-up colonoscopies (JAMA 1993;269:1278 –1281). If we extrapolate from the recently published studies that colonoscopy is the preferred screening strategy for most populations, the question must be raised again as to who will perform these colonoscopies. Thus, although it seems that colonoscopy may be the preferred screening strategy for many patients, important questions remain to be answered. JAMES D. LEWIS, M.D., M.S.C.E.
SELECTED SUMMARIES
the data and issues involved in that choice, although not addressed in our paper, were discussed by Dr. Koretz. We agree with many of Dr. Koretz’s points. First, what we and others may label as “advanced lesions” are really “surrogate markers” whose clinical significance (i.e., natural history to develop into CRC) may be speculated on but is not known. We agree that the “marginal cost-effectiveness” of colonoscopy compared with sigmoidoscopy and other screening modalities is a critically important consideration before choosing which procedure, if any, is preferred for CRC screening. As part of that assessment, safety data need to be fully considered. And we agree that the implementation of mass colonoscopy screening would need to be carefully organized because it would involve an enormous redirection of resources, direct dollar cost, and “opportunity cost” of physicians and other services which would, at least temporarily, be diverted from other health care activities. Dr. Koretz correctly points out that there are no clinical trials (which provide the strongest evidence of efficacy) to support either colonoscopic or sigmoidoscopic screening. However, we do not agree that the next step is to do a “large RCT to assess the true efficacy of avoiding having polyps left behind.” Such a trial would be lengthy, cumbersome, expensive, and is unlikely to add much, if anything, to the currently available “indirect” evidence that strongly supports the efficacy of colonoscopic screening. Such evidence comes from the RCTs of fecal occult blood test screening and the case control studies of sigmoidoscopy screening. Further, 2 clinical trials of screening sigmoidoscopy are now under way. We believe that, although additional clinical data about safety and cost-effectiveness need to be generated and analyzed, it is unnecessary to do a randomized controlled trial. Last, despite uncertainties about the details of CRC screening, we suggest that the glass should be seen as three quarters full rather than one quarter empty. A variety of screening modalities seems to be effective and cost-effective, and it seems prudent to start to implement any of them (N Engl J Med 2000;343:1641–1643). THOMAS F. IMPERIALE, M.D. DAVID F. RANSOHOFF, M.D.
WHAT LIES BEYOND THE VIEW OF THE SIGMOIDOSCOPE? Lieberman DA, Weiss DG, Bond JH, Ahnen DJ, Garewal H, Chejfec G, for the Veterans Affairs Cooperative Study Group 380 (Veterans Affairs medical centers in Portland, Oregon, Perry Point, Maryland, Minneapolis, Minnesota, Denver, Colorado, Tucson, Arizona, and Hines, Illinois). Use of colonoscopy to screen asymptomatic adults for colorectal cancer. N Engl J Med 2000;343:162–168. Imperiale TF, Wagner DR, Lin CY, Larkin GN, Rogge JD, Ransohoff DF (Department of Medicine, Indiana University Medical Center, Roudebush Veterans Affairs Medical Center, Indianapolis Gastroenterology Research Foundation, Eli Lilly, Indianapolis, Indiana; and Department of Medicine, the University of North Carolina at Chapel Hill, Chapel Hill, North Carolina). Risk of advanced proximal neoplasms in asymptomatic adults according to the distal colorectal findings. N Engl J Med 2000;343:169 –174.
1051
The two methods most commonly used to screen for colorectal cancer and adenomatous polyps are fecal occult blood testing (FOBT) and flexible sigmoidoscopy. Well-designed randomized controlled trials (Lancet 1996;348:1472–1477, 1996;348:1467–1471, N Engl J Med 1993;328:1365–1371) and case-control studies (J Natl Cancer Inst 1992;84:1572– 1575, N Engl J Med 1992;326:653– 657, Ann Intern Med 1995;123:904 –910) provide convincing evidence of the efficacy of both of these screening methods. However, both methods are less sensitive than colonoscopy for detection of colorectal neoplasms. In particular, flexible sigmoidoscopy– based screening strategies, in which only patients with neoplasms in the distal colon undergo colonoscopy, are prone to missing isolated neoplasms in the proximal colon. As a result, there is growing interest in colonoscopy as a primary screening test for average-risk patients. Two recently completed studies have addressed the potential miss rate of screening flexible sigmoidoscopy for neoplasms in the proximal colon. Lieberman et al. reported the results of a cohort study of screening colonoscopy performed at 13 Veterans Affairs medical centers (N Engl J Med 2000;343: 162–168). They enrolled 3196 patients without symptoms of lower intestinal disease, a history of colitis, colonic polyps, or colorectal cancer, and without a colonic examination (sigmoidoscopy, barium enema, or colonoscopy) within the preceding 10 years. All patients underwent colonoscopy. If the colonoscopy was not completed to the cecum, a repeat colonoscopy was attempted within 6 months of the first colonoscopy. All polyps identified at colonoscopy were removed for pathologic evaluation. Adenomatous neoplasms measuring at least 10 mm in diameter and containing villous histology, high-grade dysplasia, or invasive cancer were categorized as advanced neoplasms. The distal colon was classified as the rectum, sigmoid colon, and descending colon in the primary analysis, and as the rectum and sigmoid colon in a secondary analysis. From 4659 eligible patients, 3196 patients (96.8% men) enrolled in the study. The mean age was 62.9 years. Colonoscopy was completed to the cecum in 97.7% of the patients. Adenomatous neoplasms and advanced neoplasms were identified in 1171 patients (37.5%) and 329 patients (10.5%), respectively. Among the 128 patients with advanced proximal lesions, 62 (48.4%) had at least one adenoma distal to the splenic flexure. Using the alternative definition of the distal colon, 64 of 169 patients (37.9%) with proximal advanced neoplasms had synchronous distal adenomas. The presence of either small or large adenomas in the distal colon was positively associated with presence of advanced proximal lesions (odds ratio [OR], 2.6; 95% confidence interval [CI ], 1.7– 4.1 for small lesions; OR, 3.4, 95% CI, 1.8 – 6.5 for large lesions). Imperiale et al. reported nearly identical results in a similarly designed study (N Engl J Med 2000;343:169 –174). They performed screening colonoscopy on employees of Eli Lilly aged 40 or older with no lower gastrointestinal symptoms, no history of colonic neoplasms, and no history of inflammatory bowel disease. The distal colon was defined as the rectum, sigmoid colon, and descending colon. Advanced
1052
SELECTED SUMMARIES
neoplasms were defined as adenomas with villous features, high-grade dysplasia, or invasive cancer. Analyses were limited to subjects 50 years or older at the time of screening. A total of 1994 patients (58.9% men) were included in the analysis. The mean age was 59.8 years. Colonoscopy was completed to the cecum in 97% of patients. Advanced proximal neoplasms were identified in 50 patients (2.5%), of whom the most advanced distal finding was an advanced neoplasm in 7 (14%), a tubular adenoma in 12 (24%), a hyperplastic polyp in 8 (16%), and no polyp in 23 (46%). Compared with persons without distal polyps, advanced proximal lesions were more common in persons with distal hyperplastic polyps (OR, 2.6; 95% CI, 1.1–5.9), tubular adenomas (OR, 4.0; 95% CI, 1.9 – 8.3), and advanced neoplasms (OR, 6.7; 95% CI, 3.2– 16.6). Analyses using an alternative outcome of a proximal neoplasm at least 10 mm in diameter yielded nearly identical results. Comment. Both studies are thoughtful and well designed. Both used sound inclusion criteria and clinically meaningful outcome measures, had low rates of loss to follow-up (i.e., incomplete colonoscopy), and used appropriate statistical analytic techniques. Furthermore, the studies had nearly identical results, with more than half of all patients with advanced proximal neoplasms having no adenomas in the distal colon. Thus a logical interpretation of these studies is that, although the presence of distal colonic polyps is associated with an increased prevalence of advanced proximal colonic neoplasms, more than half of all advanced neoplasms in the proximal colon would be missed by a strategy of screening with sigmoidoscopy alone. Before we accept this conclusion, it is important to consider the potential limitations of these studies. One potential limitation is the use of the colonoscopist’s assessment of polyp location to determine the proportion of polyps that would be identified at screening flexible sigmoidoscopy. It is often difficult to state with certainty the exact anatomic location from which a polyp is removed. Given that the endoscopists were aware of the study hypothesis, this raises a small potential for bias. If the endoscopists consistently overestimated the proximal position of polyps (e.g., estimating that polyps in the descending colon were in the transverse colon), this could lead to an overestimate of the flexible sigmoidoscopy miss rate for proximal polyps. However, the consistent findings of these 2 studies (N Engl J Med 2000;343:162–168 and 2000;343:169 –174), as well as the similar findings of other studies (Gastrointest Endosc 1996;44:109 – 111 and 1999;49:727–730) suggest that this is unlikely to be the case. Another limitation of the studies is that the anatomic depth of examination by flexible sigmoidoscopy can be quite variable (Endoscopy 1999;31:227–231). Assuming that all sigmoidoscopies reach the splenic flexure may underestimate the proportion of advanced proximal neoplasms that would be missed. The secondary analysis in which the distal colon was defined as the rectum and sigmoid colon emphasizes this point (N Engl J Med 2000;343:162–168). Screening programs using FOBT and sigmoidoscopy rely on colonoscopic polypectomy for efficacy. As such, it seems inevitable that screening colonoscopy programs should produce greater reductions in cancer-related mortality than the other strategies. However, these studies were not designed to prove that screening colonoscopy is superior to flexible sigmoidoscopy every 5 years and/or annual FOBT for reducing colorectal cancer–related mortality. Because these were cross-sectional studies, we have no information on the natural
GASTROENTEROLOGY Vol. 120, No. 4
history of the lesions identified. Similarly, we do not know whether new polyps would have formed in the distal colon of persons with advanced proximal neoplasms before these became invasive cancers. These studies suggest that screening sigmoidoscopy to the splenic flexure with full colonoscopy for any patient with an adenoma of any size in the distal colon will identify approximately 80% of prevalent advanced colonic neoplasms (N Engl J Med 2000;343:162–168). Thus, the expected gain of a strategy of screening colonoscopy compared with flexible sigmoidoscopy would potentially be limited to the 20% of cases not identified with sigmoidoscopy. Combining FOBT with flexible sigmoidoscopy would be expected to identify a proportion of the advanced neoplasms missed by the screening sigmoidoscopy. In addition, some advanced lesions may be identified before they progress to invasive cancer if polyps are identified on repeat sigmoidoscopy or at colonoscopy completed for the evaluation of new symptoms. Thus, the net gain of screening colonoscopy compared with flexible sigmoidoscopy every 5 years plus annual FOBT may be relatively small, and formal comparison of these strategies in a randomized controlled trial may be impractical. In addition, the cost of obtaining this small gain in life expectancy may be quite large. Two recent cost-effective analyses both suggest that screening colonoscopy every 10 years will yield greater life expectancy at a reasonable cost per year of life gained than screening with sigmoidoscopy alone or fecal occult blood testing alone (Ann Intern Med 2000;133:573–584, JAMA 2000;284:1954 –1961). However, the analysis of Frazier et al. suggests that a strategy of annual FOBT in conjunction with sigmoidoscopy every 5 years was the most cost-effective. Furthermore, in their analysis, the combination of annual FOBT and flexible sigmoidoscopy every 5 years yields greater life expectancy than screening colonoscopy every 10 years, assuming 60% compliance with screening recommendations (JAMA 2000;284:1954 –1961). Importantly, the estimate of the most costeffective strategy was sensitive to both the compliance rate and the cost of endoscopic procedures. Because these are both moving targets, it is not possible to state with certainty whether colonoscopy or the combination of flexible sigmoidoscopy and FOBT is the most effective or the most cost-effective strategy. Certainly, compliance with annual FOBT and sigmoidoscopy every 5 years is often lower than 60% (Mor Mortal Wkly Rep CDC Surveill Summ 1999;48:116 – 121). In populations in which compliance with repeated testing is low, the long-term protection of ever having undergone colonoscopy may be much greater than ever having undergone flexible sigmoidoscopy or FOBT. Several other issues relevant to a strategy of screening colonoscopy require further attention. Firstly, how frequently should screening colonoscopy be performed? Similarly, how long can we extend the period between surveillance colonoscopies for persons with a history of adenomatous polyps? Finally, who will perform these colonoscopies? In 1993, Ransohoff and Lang estimated that a strategy of screening flexible sigmoidoscopy every 5 years would require 250,000 physician-days per year to screen all eligible persons in the United States and additional person-hours to complete the follow-up colonoscopies (JAMA 1993;269:1278 –1281). If we extrapolate from the recently published studies that colonoscopy is the preferred screening strategy for most populations, the question must be raised again as to who will perform these colonoscopies. Thus, although it seems that colonoscopy may be the preferred screening strategy for many patients, important questions remain to be answered. JAMES D. LEWIS, M.D., M.S.C.E.