Incidental Detection of Renal Cell Carcinoma is an Independent Prognostic Marker: Results of a Long-Term, Whole Population Study Helga Bjork Palsdottir, Sverrir Hardarson, Vigdis Petursdottir, Armann Jonsson, Eirikur Jonsson, Martin Ingi Sigurdsson, Gudmundur Vikar Einarsson and Tomas Gudbjartsson* From the Departments of Urology (HBP, AJ, EJ, MIS, GVE), Pathology (SH, VP) and Surgery (TG), Landspitali University Hospital and Faculty of Medicine, University of Iceland (TG), Reykjavik, Iceland Abbreviations and Acronyms CT computerized tomography ESR erythrocyte sedimentation rate IVP excretory urography RCC renal cell carcinoma Submitted for publication May 14, 2011. Study received approval from the National Bioethics Committee and Icelandic Data Protection Commission. Supported by Landspitali Research Fund and the Memorial Foundation of Bergthora Magnusdottir and Jakob Bjarnason. * Correspondence: Landspitali University Hospital, Hringbraut, IS 101 Reykjavik, Iceland (e-mail: tomasgud@landspitali.is). Purpose: The true effect of incidental detection on the survival of patients with renal cell carcinoma has been debated. We used centralized databases in Iceland to study prognostic factors of survival, focusing on the effect of incidental detection. Materials and Methods: This retrospective study included all living patients diagnosed with renal cell carcinoma in Iceland from 1971 to 2005. Hospital charts and histology were reviewed. Incidentally diagnosed renal cell carcinomas were compared to symptomatic tumors and prognostic factors were evaluated using Cox multivariate analysis. Results: Of the 910 patients 254 (27.9%) were diagnosed incidentally, most often by abdominal ultrasound (29.5%) or computerized tomography (28.3%). The incidental detection rate increased from 11.1% in 1971 through 1975 to 39.2% in 2001 through 2005 (p 0.001). During the same period the incidence increased significantly in males but in females only during the last 5 study years. Mortality remained unchanged for each gender. Incidentally detected tumors were an average of 2.6 cm smaller and diagnosed at lower stage and lower grade than symptomatic tumors. Age and histology were similar in each group. TNM stage was by far the strongest independent prognostic factor of survival but age, calendar year of diagnosis and ESR were also significant. After correcting for confounders patients with symptomatic renal cell carcinoma had worse survival than those diagnosed incidentally. Conclusions: With increased incidence and unchanged mortality the survival of patients with renal cell carcinoma has improved. This is mainly related to a steep increase in incidental detection. Incidental detection affects survival favorably and to a greater extent than can be explained by lower stage compared to the survival of patients diagnosed with symptoms. Key Words: kidney; carcinoma, renal cell; incidental findings; mortality; Iceland RENAL cell carcinoma is by far the most common malignant tumor of the kidney, accounting for 2% to 3% of all malignant tumors in northern Europe. 1,2 In most Western countries the incidence of RCC has been increasing. 3,4 This is explained by increased incidental detection, mostly due to the increased use of abdominal imaging for unrelated disease. 5,6 At the same time RCC mortality has remained stable, suggesting improved survival for patients with RCC. 1 48 www.jurology.com 0022-5347/12/1871-0048/0 Vol. 187, 48-53, January 2012 THE JOURNAL OF UROLOGY Printed in U.S.A. 2012 by AMERICAN UROLOGICAL ASSOCIATION EDUCATION AND RESEARCH, INC. DOI:10.1016/j.juro.2011.09.025
INCIDENTAL DETECTION OF RENAL CELL CANCER IS INDEPENDENT PROGNOSTIC MARKER 49 The relationship between incidental detection and incidence of RCC is not straightforward and the same applies to RCC mortality. Most studies show a similar increase in incidental detection in each gender but some only documented an increased incidence in males. 7 Also, in countries such as Denmark and Sweden, with advanced health care systems and good access to abdominal imaging, the RCC incidence has been decreasing rather than increasing. 8,9 Most studies demonstrate that patients with symptoms related to RCC have worse prognosis than those in whom RCC is incidentally detected. 7,10 17 Patard et al reported that patients with local symptoms due to RCC have better prognosis than patients with systematic symptoms. 18 Still, the role of incidental detection as an independent favorable prognostic factor for survival has been debated. 1,2,5,6,19 22 Most available studies are from single or multiple institutions rather than population based and so detection bias could possibly have affected outcomes. 7 Thus, we performed a nationwide study of a large cohort of patients who were followed a long time. The main objective was to evaluate the implications of incidental detection on RCC mortality and survival. For this we used national databases in Iceland, including a centralized RCC database containing detailed clinicopathological information. MATERIALS AND METHODS This was a retrospective, population based study of all survivors diagnosed with RCC in Iceland between January 1, 1971 and December 31, 2005. The study was approved by the National Bioethics Committee and the Icelandic Data Protection Commission. Since individuals were not identified, individual consent for the study was waived. During the study period the average population of Iceland was 250,027 individuals, that is a minimum of 204,834 on January 1, 1971 and a maximum of 299,891 on December 31, 2005 according to the Icelandic National Registry. Patients were identified separately through 2 registries, including the Icelandic Cancer Registry, and a computerized diagnosis and operation registry for hospitals in Iceland. Metanephric adenoma and oncocytoma were excluded from analysis, as were RCCs found incidentally at autopsy in patients who died of another cause. Information on the number of abdominal imaging procedures between 1986 and 2005 was obtained from all major x-ray departments in Iceland. Of 910 patients identified during the 35-year study period there were 555 males (61.0%) and 355 females with an average age of 65.2 years (range 17 to 96). Detailed clinical information, including data on patient demographics, clinical presentation and operative procedures, was gathered from hospital records. Nephrectomy was done in 712 of the 910 patients, including partial nephrectomy in 4.1% and with curative intent in 79%. Of the 910 patients 91.0% had a histologically verified diagnosis. All of these cases were reviewed by 2 experienced pathologists (SH and VP). Histological subtypes were classified according to the most recent WHO guidelines (2004), including clear cell (incorporating the multilocular subtype), papillary, chromophobe, collecting duct and unclassified RCC. The Fuhrman 4-grade scale was used to assess nuclear grade. 23 All tumors were staged according to the 2002 TNM classification system and the 2002 American Joint Committee on Cancer stage grouping. 24 Patients were divided into 2 groups, including those incidentally and those symptomatically diagnosed with RCC. Symptomatic patients were further diagnosed according to the Patard et al classification into those with localized symptoms such as hematuria, pain and flank mass, and those with systematic symptoms, such as weight loss, fever and symptoms due to metastatic lesions. 18 Incidental diagnosis was defined as tumor detected by imaging or by investigation done due to symptoms unrelated to RCC. The groups were compared for histological subtype, nuclear grade, stage, tumor size, gender and age. To investigate incidence and mortality trends the 35-year study period was divided into 7, 5-year periods. Poisson regression was used to estimate changes in incidence during the study period. Statistical calculations were done using R, version 2.11 (The R Foundation, Vienna, Austria). The Student t and Mann-Whitney U tests were used to compare continuous variables between the 2 groups based on tests of normality. The chi-square and Fisher exact tests were used to compare categorical variables. Patients were assigned a date and a cause of death or they were identified as living on December 31, 2009 using data from the Icelandic Cause of Death Register and the updated Icelandic Population Registry. Cause of death was determined from death certificates and deaths from another cause were censored. Mean followup was 74 months (range 0 to 445 months) and no patient was lost to followup. Death from RCC and disease specific (cancer specific) survival were analyzed using the Kaplan-Meier method. The log rank test was used to compare unadjusted survival curves in different groups. The Cox proportional hazards regression model was used to determine the effects of clinical and pathological parameters on mortality, including the significance of incidental detection, with p 0.05 considered statistically significant. RESULTS Of the 910 patients 656 (72.1%), including 399 males and 257 females, were diagnosed due to RCC symptoms, including 302 of 898 patients (33.6%) with localized symptoms and 342 (38.1%) with systemic symptoms. Information on symptoms was missing for 12 patients. Abdominal/flank pain in 53.0% of cases, macroscopic hematuria in 43.4%, weight loss in 26.1% and symptoms due to anemia in 18.4% were the most common symptoms. In 254 patients (27.9%), including 156 males and 98 females, RCC was diagnosed incidentally. Figure 1, A shows the incidental detection rate, which increased significantly during the study period. Table 1 lists the detection modes of incidental diagnosis. Abdominal ultrasound in 29.5% of cases and
50 INCIDENTAL DETECTION OF RENAL CELL CANCER IS INDEPENDENT PROGNOSTIC MARKER Figure 1. A, incidentally diagnosed RCC rate in males and females from 1971 to 2005 increased similarly during study period. B, abdominal imaging from 1986 to 2005 based on reports from all major institutions where imaging was done. US, ultrasound. CT in 28.3%, performed for unrelated disease or symptoms, were the most common modes. After 2000 the contribution of CT was 55.4% and that of ultrasound was 26.5%, which changed to 75% and 12.5%, respectively, by 2005 (p 0.001). Figure 1, B shows the steady increase in abdominal CT and ultrasound in Iceland from 1986 to 2005 but CT was introduced in Iceland in 1980. Ultrasound was introduced in 1986 and the number of investigations increased until year 1999 but decreased after 1999 when CT became more widely used for abdominal imaging. During the 35-year study period the age adjusted incidence of RCC was 10.5/100,000 and 6.2/100,000 for men and women, respectively, and it increased significantly with each calendar year (p 0.001). This increase was mostly due to the number of men, in whom the age adjusted incidence rose from 7.6/ 100,000 in 1971 through 1975 to 13.2/100,000 in 2001 through 2005 (p 0.001, fig. 2). Corresponding values for women were 6.5/100,000 and 8.2/100,000 (p 0.24, fig. 2). RCC mortality remained relatively unchanged for each gender during the study period (fig. 2). With an increasing incidence and relatively stable mortality the survival of patients with RCC has improved in Iceland since 1971, particularly for men but also for women. Table 2 lists incidentally diagnosed RCCs and those diagnosed by symptoms. Incidentally detected tumors were an average of 2.6 cm smaller (p 0.001) but age, laterality and gender ratio were comparable Table 1. Mode of detecting incidentally detected RCCs in Iceland from 1971 to 2005 No. Pts (%) Abdominal ultrasound 75 (29.5) CT 72 (28.3) IVP 56 (22.0) Laparotomy 9 (3.5) Chest x-ray 7 (2.8) Aortic angiography 4 (1.6) Other 31 (12.2) in the 2 groups. A significant decrease in tumor size was seen during the study period (p 0.001), including the proportion of tumors larger than 7 cm. Table 2 shows information on Fuhrman nuclear grade and TNM stage by gender. Trends were similar for men and women. Incidentally detected tumors were diagnosed at a lower TNM stage than symptomatic tumors, for example 57.1% and 10.2% for stages I and IV compared to 18.8% and 43.7%, respectively, for symptomatic RCC (p 0.001). Incidentally detected tumors were also diagnosed at a lower Fuhrman grade than symptomatic RCC with grade 1 or 2 in 75.6% vs 48.4% for symptomatic disease (p 0.001). Grade and TNM stage decreased significantly during the study period. A strong correlation was found for grade and stage with linear regression showing an increase of 0.26 in grade for a unit increase in stage (p 0.001). Histological subtypes were comparable for each group and similar for men and women (table 2). Clear cell carcinoma was most common in each groups at 89.4%, followed by papillary RCC at 8.0% and chromophobe RCC at 1.8%. Mean hemoglobin was significantly higher in the incidental than in the symptomatic group (138.5 vs 127.0 gm/l) and ESR was lower (25.2 vs 45.3 mm per hour). Figure 3 shows estimated disease specific survival for the whole group and for stages I to IV. Five-year survival for each stage was 91.9,% 80.1,% 61.8% and 11.6%, respectively. An improvement in survival was seen in the latter part of the study period, especially for the last 5-year period when survival attained 71.0% vs 43.5% between 1971 and 1975 (p 0.005). Table 3 lists significant clinicopathological prognostic factors of disease specific death (due to RCC). On univariate analysis histological subtype was a significant prognostic factor of death with a better prognosis for chromophobe RCC (HR 0.19) than for papillary (HR 0.65) and clear cell (HR 1) RCC (p 0.001). The same was true for increasing tu-
INCIDENTAL DETECTION OF RENAL CELL CANCER IS INDEPENDENT PROGNOSTIC MARKER 51 factors were age (HR 1.02) and high ESR (HR 1.004). However, recent calendar year of diagnosis was associated with improved survival (HR 0.98). After correcting for age, diagnosis year, ESR, hemoglobin, nuclear grade and TNM stage symptomatic patients with RCC had significantly worse survival than those diagnosed incidentally (symptomatic HR 1.40, 95% CI 1.01 1.93, p 0.043). However, this was only true for patients with systematic and not localized symptoms (HR 1.47, CI 1.03 2.09, p 0.032 vs HR 1.31, CI 0.92 1.85, p 0.13). Figure 2. Age adjusted incidence and mortality in patients diagnosed with RCC from 1971 to 2005. There was significant increase in incidence in males from 1971 to 2005 but only in females from 2001 to 2005. Mortality was relatively unchanged during whole period for each gender. mor size (HR 1.013), tumors on the right side (HR 1.36), renal vein extension (HR 1.99) and lower hemoglobin (HR 0.98). Histological grades 3 (HR 4.3) and 4 (HR 8.6) were significantly associated with worse prognosis compared to grade 1 (HR 1.0). Patients with symptomatic diagnosis (HR 3.41) also had worse prognosis than those in whom RCC was detected incidentally. Also, compared with asymptomatic patients (HR 1.0) those with localized (HR 2.16) and systematic (HR 5.10) symptoms had increased mortality. On Cox multivariate analysis TNM stage was by far the strongest independent prognostic factor of mortality with an HR of 17.7 for stage IV vs I disease (p 0.001). Other independent negative prognostic DISCUSSION Incidental detection of RCC affects survival favorably. This is a survival benefit that is greater than can be explained by lower stage or grade, or differences in patient demographics. The survival of patients with RCC in a whole nation has improved with an increased incidence and unchanged mortality. The main reason for this survival benefit appears to be a steep increase in incidental detection, as reflected in the increased use of abdominal ultrasound and CT reported from all institutions where imaging is done in Iceland. Table 4 shows that the rate of incidentally diagnosed RCC in recently published studies is in the 30% to 50% range but the rate often exceeds 50%, as in our cohort after 2000. Similar results were previously published but no group studied a population based cohort, 5,6 as we did. The survival benefit of an incidental diagnosis was slightly higher in the last 10 years compared to the first 25 years of the study period. This Table 2. Clinicopathological parameters of patients diagnosed with symptoms due to RCC and those with incidentally detected RCC in Iceland from 1971 to 2005 Incidental RCC Symptomatic RCC All RCCs p Value No. pts (%) 254 (27.9) 656 (72.1) 910 Mean age 65.7 65.0 65.2 0.45 Male/female ratio 1.6 1.6 1.6 0.93 No. rt tumors (%) 145 (57.3) 324 (50.6) 469 (52.5) 0.35 Mean hemoglobin (gm/l) 138.5 127.0 130.2 0.001 Mean ESR (mm/hr) 25.2 45.3 39.8 0.001 Mean tumor size (cm) 5.5 8.1 7.3 0.001 No. nuclear grade (%): 0.001 1 29 (12.2) 22 (3.8) 51 (6.2) 2 151 (63.4) 261 (44.6) 412 (50.1) 3 48 (20.2) 227 (38.8) 275 (33.4) 4 10 (4.2) 75 (12.8) 85 (10.3) No. TNM stage (%): 0.001 I 145 (57.1) 123 (18.8) 268 (29.5) II 29 (11.4) 84 (12.8) 113 (12.4) III 54 (21.3) 161 (24.6) 215 (23.7) IV 26 (10.2) 286 (43.7) 312 (34.4) No. histology (%): 0.08 Clear cell 213 (89.1) 527 (89.5) 740 (89.4) Papillary 23 (9.6) 43 (7.3) 66 (8.0) Chromophobe 1 (0.4) 14 (2.4) 15 (1.8)
52 INCIDENTAL DETECTION OF RENAL CELL CANCER IS INDEPENDENT PROGNOSTIC MARKER Figure 3. Estimated disease specific survival from 1971 to 2005. A, patients diagnosed with symptoms due to RCC and those diagnosed incidentally. B, TNM stage of patients diagnosed with RCC. Table 3. Cox multivariate analysis of factors significantly prognostic for disease specific mortality in patients with RCC diagnosed in Iceland from 1971 to 2005 HR (95% CI) p Value Age 1.017 (1.008 1.026) 0.001 Diagnosis yr 0.983 (0.972 0.994) 0.002 ESR 1.004 (1.001 1.008) 0.03 TNM stage: 0.001 I 1 (referent) II 2.76 (1.66 4.58) III 4.26 (2.73 6.63) IV 17.65 (11.29 27.59) Symptomatic diagnosis 1.40 (1.01 1.93) 0.04 may have been due to the greater sensitivity of CT and ultrasound for diagnosing RCC compared to IVP. The implication on survival of incidentally detected tumors is not obvious and there is disagreement in the literature about its significance. Some groups have reported it as an independent positive prognostic factor while others have not. 2,7,19,20,25 The increased number of cases, particularly incidentally diagnosed cases, together with longer followup allowed us to extend our earlier observations of variables that may be prognostic of increased survival. 7 As previously reported, TNM staging of RCC, patient age, diagnosis year and ESR were significant, independent prognostic markers of survival but nuclear grade was not. However, while correcting for these factors on multivariate analysis, we also found that a symptomatic diagnosis of RCC carried a significantly higher risk of death than an incidental diagnosis (HR 1.40, 95% CI 1.01 1.93). As noted by Patard et al. 18 this difference is mostly due to patients with systematic RCC symptoms rather than localized symptoms. An epidemiological variable that differs between incidentally and symptomatically diagnosed patients may explain our results. An example of such a variable might be a different frequency of interaction with health care between the 2 groups. It might also be due to possible lead time and stage migration bias, in that tumors discovered earlier may be less invasive and have lower malignant potential, resulting in more favorable prognosis. An alternative and intriguing explanation would be a biological difference between incidentally and symptomatically diagnosed RCCs. Subgrouping RCC into more biological subclasses than the current TNM classification system allows may explain our observation. Such classification could be based on emerging molecular techniques, such as classifying genes based on gene expression profiles or examination of the mutation spectra of genes involved in RCC pathogenesis. Future research must resolve these important issues. 26 By reviewing all RCCs in the whole of the Icelandic population, including nonoperated patients, we may have decreased certain types of ascertainment bias. To our knowledge the current series and our Table 4. Incidental detection in recent major studies of RCC References (study period) No. Incidental/Total No. (%) Luciani et al: 27 1982 1983 10/77 (13) 1996 1997 90/152 (59) Tsui et al 17 (1987 1998) 95/633 (15) Patard et al 2 (1984 1999) 151/400 (38) Lee et al 20 (1989 1997) 411/721 (57) Ficarra et al 21 (1976 2000) 630/1,446 (44) Ou et al 28 (1982 2001) 80/312 (26) Ghalayini and Bani-Hani 29 (1992 2001) 31/119 (26) Gudbjartsson et al 7 (1971 2000) 172/701 (25) Kawaciuk et al 30 (1982 2001) 135/396 (34) Present series (1971 2005) 254/910 (28)
INCIDENTAL DETECTION OF RENAL CELL CANCER IS INDEPENDENT PROGNOSTIC MARKER 53 previously published study 7 are the only population based RCC studies to have analyzed the effect of incidental detection on survival. 7 A further strength of the current study was the reclassification and verification of diagnosis of all cases by only 2 pathologists. In conclusion, incidental detection appears to affect survival favorably and to a greater extent than can be explained by differences in stage, grade or patient demographics compared to those of patients with symptoms. REFERENCES 1. Parkin DM, Bray F, Ferlay J et al: Estimating the world cancer burden: Globocan 2000. Int J Cancer 2001; 94: 153. 2. Patard JJ, Rodriguez A, Rioux-Leclercq N et al: Prognostic significance of the mode of detection in renal tumours. BJU Int 2002; 90: 358. 3. Chow WH, Devesa SS, Warren JL et al: Rising incidence of renal cell cancer in the United States. JAMA 1999; 281: 1628. 4. Bretheau D, Koutani A, Lechevallier E et al: A French national epidemiologic survey on renal cell carcinoma. Cancer 1998; 82: 538. 5. Pantuck AJ, Zisman A and Belldegrun AS: The changing natural history of renal cell carcinoma. J Urol 2001; 166: 1611. 6. Bos SD, Mellema CT and Mensink HJ: Increase in incidental renal cell carcinoma in the northern part of the Netherlands. Eur Urol 2000; 37: 267. 7. Gudbjartsson T, Thoroddsen A, Petursdottir V et al: Effect of incidental detection for survival of patients with renal cell carcinoma: results of population-based study of 701 patients. Urology 2005; 66: 1186. 8. Mathew A, Devesa SS, Fraumeni JF Jr et al: Global increases in kidney cancer incidence, 1973 1992. Eur J Cancer Prev 2002; 11: 171. 9. Ferlay J, Parkin DM and Steliarova-Foucher E: Estimates of cancer incidence and mortality in Europe in 2008. Eur J Cancer 2010; 46: 765. 10. Katusin D, Uzarevic B, Petrovecki M et al: Clinical, histopathological and flow-cytometric properties of incidental renal cell carcinomas. Urol Res 2000; 28: 52. 11. Konnak JW and Grossman HB: Renal cell carcinoma as an incidental finding. J Urol 1985; 134: 1094. 12. Thompson IM and Peek M: Improvement in survival of patients with renal cell carcinoma the role of the serendipitously detected tumor. J Urol 1988; 140: 487. 13. Ueda T, Yasumasu T, Uozumi J et al: Comparison of clinical and pathological characteristics in incidentally detected and suspected renal carcinoma. Br J Urol 1991; 68: 470. 14. Tsukamoto T, Kumamoto Y, Yamazaki K et al: Clinical analysis of incidentally found renal cell carcinomas. Eur Urol 1991; 19: 109. 15. Rodriguez-Rubio FI, Diez-Caballero F, Martin-Marquina A et al: Incidentally detected renal cell carcinoma. Br J Urol 1996; 78: 29. 16. Sweeney JP, Thornhill JA, Graiger R et al: Incidentally detected renal cell carcinoma: pathological features, survival trends and implications for treatment. Br J Urol 1996; 78: 351. 17. Tsui KH, Shvarts O, Smith RB et al: Renal cell carcinoma: prognostic significance of incidentally detected tumors. J Urol 2000; 163: 426. 18. Patard JJ, Leray E, Cindolo L et al: Multiinstitutional validation of a symptom based classification for renal cell carcinoma. J Urol 2004; 172: 858. 19. Bretheau D, Lechevallier E, Eghazarian C et al: Prognostic significance of incidental renal cell carcinoma. Eur Urol 1995; 27: 319. 20. Lee CT, Katz J, Fearn PA et al: Mode of presentation of renal cell carcinoma provides prognostic information. Urol Oncol 2002; 7: 135. 21. Ficarra V, Prayer-Galetti T, Novella G et al: Incidental detection beyond pathological factors as prognostic predictor of renal cell carcinoma. Eur Urol 2003; 43: 663. 22. Harada K, Sakai I, Ishimura T et al: Clinical symptoms in localized renal cell carcinoma reflect its invasive potential: comparative study between incidentally detected and symptomatic diseases. Urol Oncol 2006; 24: 201. 23. Fuhrman SA, Lasky LC and Limas C: Prognostic significance of morphologic parameters in renal cell carcinoma. Am J Surg Pathol 1982; 6: 655. 24. Greene FL, Page DL, Fleming ID et al: AJCC Cancer Staging Manual, 6th ed. Chicago: American Joint Committee on Cancer 2002. 25. Parsons JK, Schoenberg MS and Carter HB: Incidental renal tumors: casting doubt on the efficacy of early intervention. Urology 2001; 57: 1013. 26. Brannon AR, Reddy A, Seiler M et al: Molecular stratification of clear cell renal cell carcinoma by consensus clustering reveals distinct subtypes and survival patterns. Genes Cancer 2010; 1: 152. 27. Luciani LG, Cestari R and Tallarigo C: Incidental renal cell carcinoma age and stage characterization and clinical implications: study of 1092 patients (1982 1997). Urology 2000; 56: 58. 28. Ou YC, Yang CR, Ho HC et al: The symptoms of renal cell carcinoma related to patients survival. J Chin Med Assoc 2003; 66: 537. 29. Ghalayini IF and Bani-Hani IH: Detection, staging and clinical implications of renal cell carcinoma. Saudi Med J 2003; 24: 79. 30. Kawaciuk I, Hyrsl L, Dusek P et al: Influence of tumour-associated symptoms on the prognosis of patients with renal cell carcinoma. Scand J Urol Nephrol 2008; 42: 406.