Elevated Serum Vitamin B12 Levels Associated With CRP as a Predictive Factor of Mortality in Palliative Care Cancer Patients

Elevated Serum Vitamin B12 Levels Associated With CRP as a Predictive Factor of Mortality in Palliative Care Cancer Patients

Vol. 20 No. 2 August 2000 Journal of Pain and Symptom Management 93 Original Article Elevated Serum Vitamin B12 Levels Associated With CRP as a Pr...

301KB Sizes 0 Downloads 31 Views

Vol. 20 No. 2 August 2000

Journal of Pain and Symptom Management

93

Original Article

Elevated Serum Vitamin B12 Levels Associated With CRP as a Predictive Factor of Mortality in Palliative Care Cancer Patients: A Prospective Study Over Five Years Philippe Geissbühler, MD, Bernadette Mermillod, BSc, and Charles-Henri Rapin, MD Département de Gériatrie (P.G., C.-H. R), and Division d’Informatique Médicale (B.M.), Hôpitaux Universitaires de Genève, Genève, Switzerland; Center for Bioethics (C.-H.R.), IRCM, University of Montreal, Montreal, Canada; and Centre Intefacultaire de Gérontologie (C.-H.R.), University of Geneva, Geneva, Switzerland

Abstract The relationship between vitamin B12 levels and survival was studied in a group of 161 terminally ill cancer patients who were recruited consecutively between 1988 and 1989. Their average age was 74.7 years. The length of survival decreased with the increase in serum vitamin B12 levels (P ⫽ 0.0015, Cox model). In multivariate analyses, C-reactive protein (CRP) was the most important prognostic factor in this population, and vitamin B12 provided information independent of CRP in predicting survival. These data indicate that an elevated serum vitamin B12 level is a predictive factor for mortality in patients with cancer, independent of CRP or other factors. Multiplying it by the CRP makes it possible to create a new, easy-to-use prognostic index, which can distinguish different levels of mortality risk at three months. J Pain Symptom Manage 2000;20:93–103. © U.S. Cancer Pain Relief Committee, 2000. Key Words Vitamin B12, CRP, palliative care, survival, prognostic factor

Introduction The search for prognostic factors that would make it possible to estimate quite precisely the survival time of a patient in the terminal phase of cancer has always been a point of interest in palliative care.1 It would open the door to better treatment strategies by providing an additional decision-making tool; for example,

Address reprint requests to: Charles-Henri Rapin, MD, Department of Geriatrics, Hospital of the University of Geneva, 35 rue des Bains, CH-1205 Geneva, Switzerland. Accepted for publication: September 11, 1999. © U.S. Cancer Pain Relief Committee, 2000 Published by Elsevier, New York, New York

whether to continue or discontinue chemotherapy and further investigation. Ideally, it would prevent futile overtreatment. Many types of prognostic indicators are now being used and discussed: Functional scales (Karnofsky, Eastern Cooperative Oncology Group (ECOG), combinations of certain symptoms, psychological assessments), combinations of biological and nutritional factors, and these two types of indicators combined.2 In an unpublished preliminary study testing the prognostic value of certain biological/nutritional parameters in elderly hospitalized subjects, we discovered an apparently paradoxical result: an elevated vitamin B12 level predicted 0885-3924/00/$–see front matter PII S0885-3924(00)00169-X

94

Geissbühler et al.

early death in this population. In this study, which was conducted on a small group of patients who had undergone many laboratory tests, a false positive was possible. We wanted to check, in a prospective study, the relationship between an elevated serum B12 and an earlier death, and if that were the case, if it predicted death independently of the Prognostic Inflammatory and Nutritional Index (PINI),3 a prognostic index then used in our institution, and other factors known to elevate vitamin B12 levels, including the presence of liver problems (particularly liver metastasis), leukemia, or myelodysplastic syndrome.

Vol. 20 No. 2 August 2000

Table 1 Patients With One Tumor, Type of Tumor Type of tumor

Number Liver metastasis (%)

Lung Digestive tract Prostate Breast Hematopoietic system Kidney and urinary tract Gynecological ENT Nervous system Melanoma Liver Pancreas Mesothelioma Undetermined Two cancers Total

24 29 13 25 14 8 9 7 4 2 4 7 1 3 11 161

8 19 1 7 2 4 1 1 0 1 — 2 0 1 5 52

33 66 8 28 14 50 11 14 0 50 — 29 0 33 45 32

Methods This was a prospective study whose protocol was approved by the ethics committee. In this study, we observed the survival length of all patients with cancer who were admitted consecutively between September 15, 1988 and November 27, 1989 to the Centre de soins continus (CESCO) in Collonge-Bellerive. This center admits patients who may particularly benefit from palliative care; it is a geriatric facility attached to the university institutions in Geneva (Institutions Universitaires de Gériatrie de Genève). We excluded from the study those patients who had received vitamin B12 supplements in the weeks preceding their admission to CESCO.

Clinical Data Shortly after the patient left the facility, we noted, by consulting the patient’s medical file, the type of tumor the patient had; the presence or site of the metastasis, particularly the liver; the type of examination performed to detect liver metastasis (autopsy, computerized tomography, or ultrasonography); and the presence of liver problems that were not tumorrelated (Tables 1 and 2). The date of admission into the study was considered the date of the blood test performed when the patient was admitted to CESCO. The dates of death were confirmed by consulting the civil status registry in the canton of Geneva.

Laboratory Examinations We selected laboratory variables that might be linked to the pathophysiology of an ele-

vated serum vitamin B12 (folates, leukocytes, liver function tests, creatinine), those that entered into the calculation of the PINI prognostic index (the C-reactive protein [CRP], the ␣lacid glycoprotein [AGP], the albumin, and the prealbumin), and the hemoglobin, which has been studied as a prognostic factor in geriatrics.4 A fasting blood test is done routinely on all patients admitted to CESCO, the day following admission. A serum sample was frozen and sent to the central clinical chemistry laboratory (Geneva Canton University Hospital) for analyses of vitamin B12 and folates using the RIA technique for both parameters simultaneously, after alkaline denaturation of the samples (Ref. Dualcont solid phase no boil, Diagnostic Products Corporation, Los Angeles, CA, USA, #KDSP2). The C-reactive protein, the ␣l-acid glycoprotein (AGP), the albumin, and prealbumin were determined by nephelometry in the Table 2 Patients With 2 Tumors, Types of Tumors (n ⫽ 11) Tumor 1 Lung Lung Lung Colon Colon Colon Colon Prostate Prostate Breast Melanoma

Tumor 2

Liver Metastases

Bladder Bladder Larynx Intestine Breast Breast Prostate Pancreas Stomach Uterus Waldenström

no no no yes yes yes no no no yes yes

Vol. 20 No. 2 August 2000

Elevated Vitamin B12 Predicts Mortality

CESCO laboratory (Array protein system, Beckmann). The alkaline phosphatase, bilirubin, aspartame aminotransferase (AST), alanine aminotransferase (ALT), and creatinine were assayed on Technicon RAXT in the clinical chemistry laboratory of the university hospital for geriatrics in Geneva. The alkaline phosphatase was measured using the kinetic method with a 2-amino-2-methyl-1-propanolol buffer; the AST and ALT were also measured by the kinetic method with pyridoxal phosphate activation at 37⬚C; the creatinine was measured using a picric acid colorimetric procedure in an alkaline medium. Finally, the leukocytes were counted in the CESCO laboratory by resistance detection, and the hemoglobin was determined using a variant of the cyanmethemoglobin method with the photometric measurement at 540 nm. These two last measurements were done on SYSMEX CC-180. We calculated the PINI prognostic index (Prognostic Inflammatory and Nutritional Index): PINI = AGP [ mg/1 ] × CRP [ mg/1 ] --------------------------------------------------------------------------------------------------Albumin [ g/1 ] × Pre-albumin [ mg/1 ] A certain number of blood samples were not completed in a satisfactory manner. Table 3 lists the number of missing values for each laboratory examination. We chose not to exclude a posteriori the four patients from whom we did not succeed in obtaining serum vitamin B12 values, as the samples were not sent to the laboratory. They belonged to a 72-year-old woman with metastatic breast cancer who had survived for nine days, a 76-year-old man with metastatic prostate cancer who had survived for 173 days, a 76-year-old woman suffering from lymphoma who had survived for 173 days, and a 62-year-old woman with breast cancer and a tumor of the colon with liver metastasis, who had survived for 31 days.

Classification of Variables Three binary variables (yes/no), liver problem, liver metastasis, and hematologic malignancy, were created to explore clinical disorders that could influence vitamin B12 metabolism. The liver problem variable included any diagnosis that involved the liver, for example, alcoholic cirrhosis of the liver or viral hepatitis. The hematologic malignancy vari-

95

Table 3 Missing Tests Variables Bilirubin ALT AST Folates Leukocytes Vitamin B12 Hemoglobin Creatinine Alkaline phosphatase C-reactive protein (CRP) ␣1-acid glycoprotein acid (AGP) Albumin Pre-albumin

Number 65 36 25 8 5 4 3 2 1 1 1 0 0

able included the lymphomas, leukemias, the myelomas, and myelodysplastic syndromes. It was necessary to define the liver problem variable because too many of the bilirubin, AST, and ALT values were missing. These latter were not kept as we noted that they selected a subgroup of patients not comparable to the rest of the group. Vitamin B12 and other ongoing variables (PINI, CRP) were categorized in three groups of the same size for graphical display of survival curves.

Statistical Methods Survival was measured from day one of the study until death (from any cause) or the date of the last contact. The only patient censored was known to be alive at the cut-off date, which was March 21, 1994. Survival probability was calculated using the Kaplan-Meier technique.5 Confidence intervals (CI) for survival probabilities at a given time were calculated according to Dorey and Korn,6 using the RothmanWilson upper limit7 and the Simon-Lee lower limit.8 Reflected confidence intervals for median survival were calculated.9 The survival curves were compared using the Mantel-Cox10 log-rank test, or trend test, when three ordered groups were compared.11 Multivariate analysis using the Cox proportional hazards model was also performed.12 Some very skewed variables (i.e., B12, CRP, prealbumin, and PINI) were log-transformed.

Results A total of 161 patients, 74 men and 87 women, were recruited. Their ages ranged from 46 to 96 years, the average was 74.7 (me-

96

Geissbühler et al.

dian 76 years, SD 9.8 years). From 1988 to 1994, 160 patients died. The only surviving patient, a man born in 1904, had a survival time of five years and 2 months (the longest in the group) and suffered from colon cancer with no metastasis to the liver. The global median survival time was 45 days (CI 95%: 32–56 days). The shape of the global survival curve (Fig. 1) reflects the type of population monitored at CESCO, with a high mortality rate in the first weeks of monitoring. The highest mortality corresponded to the highest B12 levels. It decreased parallel to the decrease in B12 levels, in significant fashion (Cox model, P ⫽ 0.0015, Figs. 2a and 2b). In the univariate analysis, PINI and its four components, CRP, AGP, albumin, and prealbumin, were predictors of the length of survival, as were other variables (i.e., B12, leukocytes, liver problems, and age [Table 4]). Of all these variables, CRP and PINI had the greatest and same power of discrimination. For this reason, CRP was chosen. None of the other PINI components gave any more information in addition to CRP. The same was true for PINI: all the information in PINI was already present in CRP. B12 remained the only variable containing information not already included in CRP. Once CRP and B12 were included in the model, no

Fig 1.

Vol. 20 No. 2 August 2000

variable retained independent information. Given the great similarity between the two coefficients (Table 5), a simplified model may be obtained by taking the product of the two variables, which we called BCI (B12 CRP index). The BCI was easier to use in the calculations. For example, a BCI ⬎ 40,000 was associated with a three-month mortality of 90%, compared to 50% with a BCI ⬍ 10,000 (Figs. 3a and 3b). The link between the BCI and survival is clearly significant (trend test P ⬍ 0.0001) and confirms the Cox model obtained. Another graphic representation showing the link between B12, CRP, and survival is presented in Figure 4. This figure gives a scatterplot of CRP and B12, with observations denoted by circles, the area of which are proportional to length of survival. The two lines correspond to BCI (CRP multiplied by B12) of 10,000 for the lower one, and 40,000 for the upper one. As areas of circles are proportional to the length of survival, we see that patients with a long survival are for the majority located in the left lower third of the figure. As we move to the right upper corner, the circles become smaller and smaller. Of course, some short survival can be seen in the left lower corner and some long ones in the middle of the figure. There is a significant link between the presence of metastasis, a tumor or liver problem,

Overall survival, first year. Vertical lines correspond to 95% confidence intervals.

Vol. 20 No. 2 August 2000

Fig. 2a.

Elevated Vitamin B12 Predicts Mortality

Survival according to B12, first year. Vertical lines correspond to 95% confidence interval.

Fig. 2b.

Survival according to B12, five years. Trend test ⫽6.066 (df⫽1) p ⫽ .014.

97

98

Geissbühler et al.

Vol. 20 No. 2 August 2000

Table 4 Variables Tested in the Cox Model Variables Hemoglobin Creatinine Leukocytes PINI CRP AGP Albumin Pre-albumin Liver problem Hematologic malignancy Age Sex B12

Table 5 Cox Model Findings P-value

0.41 0.41 0.001 ⬍0.00005 ⬍0.00005 0.0005 0.002 0.004 0.008 0.15 0.02 0.37 0.0015

These results are the univariate P-values of the Cox model. All these variables were considered for multivariate analysis.

and an elevated vitamin B12 level (P ⬍ 0.001). There is also a strong link between the presence of a liver problem and an elevated CRP (P ⬍ 0.001). We defined a hematologic malignancy variable corresponding to the total number of cases of lymphoma (n ⫽ 9), leukemia (n ⫽ 1), myelodysplastic syndrome (n ⫽ l), and multiple myeloma (n ⫽ 3), because of the possible link between an elevated leukocyte count and an elevated B12. Comparing this group to the other patients, there was no significant difference in the B12 value (Mann-Whitney, P ⫽ 0.32). For all patients, the correlation between B12 and leukocytes was low (Spearman, P ⫽ 0.18).

Discussion This study shows that there is a significant link between an elevated serum B12 level and mortality in elderly patients suffering from cancer in its advanced stage. This link was already noted by Carmel and Eisenberg in 1977.13 In their study, a group of 24 patients with a high serum B12 level had a median survival time of one month, independent of associated transcobalamine levels. This contrasted significantly with the four-month median survival in 115 patients with normal or low B12 levels. They also noted a correlation between an elevated B12 and metastasis to the liver or elsewhere. No pathophysiological explanation for this finding could be validated in this study. Our findings concur with those of Carmel and Eisenberg in the survival differences observed in relation to the BCI, the prognostic in-

Variable log(B12)b log(CRP)b

Coefficient

Hazard Ratioa (95% CI)

0.61 0.68

1.20 (1.03–1.41) 1.23 (1.11–1.36)

aHazard blog

ratio is given for doubling B12 or CRP. to the base 10.

dex created from the Cox model that combines CRP and B12 values. We also noted an association between an elevated vitamin B12 and the presence of liver metastasis or a liver problem. However, we did not demonstrate any unfavorable link between an elevated B12 and hematologic malignancy, in the sense in which we defined it earlier, or leukocytosis. The group of patients suffering from hematologic malignancy presented an even better survival time than the others. An elevated vitamin B12 level in myeloproliferative diseases, notably in chronic myeloid leukemia, Vaquez’s disease, hepatocellular carcinoma, liver diseases, and inflammatory diseases, has long been recognized.14,15 It is secondary to that of cobalamin carrier proteins.16–18 These proteins are of two types: transcobalamin II and the R-binders. A recent review details the process of transcobalamin metabolism.19 The association between elevated circulating levels of transcobalamins, and neoplastic or inflammatory diseases, suggests that these proteins can behave like an “acute phase reactant.” It must be noted that in these cases, the B12 concentrations remain normal, unless there is a coexisting liver problem, thus causing an elevation in B12 levels. One can readily imagine a pathophysiological model in which an inflammatory condition associated with a liver ailment leads to an elevation in serum B12 levels. The inflammatory condition, the presence of necrosis, pus, or an increase in cellular turnover would lead to an increase in transcobalamine synthesis. The hepatic clearance would be decreased, with the liver even contributing to the increase in serum B12 levels by releasing transcobalamins and B12. Problems with vitamin B12 metabolism, when there is liver involvement 20,21 elevated transcobalamins as an indicator of inflammatory activity in rheumatoid polyarthritis,22–24 and the use of B12 levels as a prognostic factor in multiple myeloma,25 have been the subject

Vol. 20 No. 2 August 2000

Fig. 3a.

Elevated Vitamin B12 Predicts Mortality

Survival according to BCI, first year. Vertical lines correspond to 95% confidence interval.

Fig. 3b.

Survival according to BCI, five years. Trend test ⫽ 23.75 (df⫽1) p ⬍ 0.0001.

99

100

Geissbühler et al.

Vol. 20 No. 2 August 2000

Fig. 4. Relation between B12, CRP, and Survival. This figure gives a scatterplot of CRP and B12, with observations denoted by circles with area proportional to length of survival. The two lines represent a BCI (CRP multiplied by B12) of 10,000 for the lower one and 40,000 for the upper one. The gray circle represents the only survivor.

of clinical studies. Finally, the increased transcobalamin synthesis by certain tumors is in the process of being evaluated.26–28 The CRP, however, is a typical “acute phase protein,” whose metabolism is better known, and whose links with cytokines, particularly interleukin-I (IL-I) and IL-6, have been clearly established.29 Its role as a prognostic factor30 or indicator of inflammation in various pathologies has also been the subject of clinical studies, notably in those that have elevated transcobalamins.31,32 The extremely complex interplay of cytokines, which is where the “acute phase response” begins,33 causes regulatory mechanisms to intervene; these differ depending on the pathology or tissue where it

takes place. This complexity makes a different expression of “acute phase proteins” plausible, depending on the inflammatory context, and we can suppose that it could lead to an elevation in the CRP and/or vitamin B12, independent of one another. To summarize, an elevated B12 would be an indicator of hepatic involvement and inflammation. We postulate that it reflects pathophysiological processes different from those involving the CRP, which would explain their independence in predicting survival. The role of B12 as a prognostic factor in palliative care must naturally be validated by others or in a new set of patients, but on the basis of this study, we can already select B12 as a candidate when looking

Vol. 20 No. 2 August 2000

Elevated Vitamin B12 Predicts Mortality

Fig. 5.

101

Relationship between PINI and CRP, r ⫽ 0.96.

for prognostic factors by combining data from the clinical chemistry laboratory. The PINI is a validated prognostic index which shows a good correlation with survival. A bad prognosis is associated with PINI values higher than 25, indicating high mortality at two months.3,34,35 Nevertheless, we propose to replace the PINI measurement with the CRP measurement, given the interdependence of albumin, prealbumin, alpha-1 globulin, and CRP values; the equivalence of prediction given by CRP and B12, or PINI and B12; and the more simple model with CRP and the strong correlation between CRP and PINI (r ⫽ 0.96, Figure 5). Put in other words, PINI may just be a more expensive and complicated measure of CRP. In rejecting PINI, one must not, however, set aside the albumin and prealbumin readings.

While their levels are, of course, correlated to CRP levels, they nonetheless remain indicators of nutritional conditions, particularly when they do not develop “in concert,” as for example, when a low CRP is associated with a low albumin in a given patient. We have not included performance status rating (e.g., Zubrod or Karnovsky Performance status), which is a limitation of the strength of this study, because they were not routinely performed at this time in our institution. This could be the object of a complementary study. This study has enabled us to rediscuss the place given to prognostic indicators in our practice. A recent study36 shows that experienced clinicians are good predictors of the length of survival in the terminal phase of cancer. In our hospitals, however, it is young physi-

102

Geissbühler et al.

cians who are most often confronted with terminal situations. They often lack training and support when faced with the “death-failure” association, and sometimes they will embark on the futile and very costly process of overdiagnosis and overtreatment.37–40 The situation could be improved. In the institutional setting, death is rarely an unexpected event; the probability of its occurrence in a given patient may well be envisaged. This notion of making a prognosis on death is indeed a didactic element that may help young professionals to envisage death, not as an unexpected failure, but as the natural progress of the patient’s disease. In conclusion, if simple laboratory tests such as those suggested in this study provide additional elements that make our interventions more appropriate, we must not reject them. Of course, like any other test, they must be part of an overall perspective on the patient. Any tool that gives us a better grasp of the patient’s situation will help us deal with ambivalent feelings induced by the proximity of death.

References 1. Den Dass N. Estimating length of survival in end-stage cancer: a review of the literature. J Pain Symptom Manage 1995;10:548–555. 2. Maltoni M, Pirovano M, Nanni O, Labianca R, Amadori D. Facteurs pronostiques chez les patients en phase terminale. Eur J Palliat Care 1994;1:122– 125. 3. Ingengleck Y, Carpentier A. A prognostic inflammatory and nutritional index scoring critically ill patients. Int J Vitam Nutr Res 1985;55:91–101. 4. Verdery RB, Goldberg AP. Hypocholesterolemia as a predictor of death: a prospective study of 224 nursing home residents. J Gerontol 1991;46: M84–M90. 5. Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. JASA 1958;53:457– 481. 6. Dorey FJ, Korn EL. Effective sample sizes for confidence intervals for survival probabilities. Stat Med 1987;6:679–687. 7. Rothman KJ. Estimation of confidence limits for the cumulative probability of survival in life table analysis. J Chronic Dis 1978;31:557–560. 8. Simon R, Lee YJ. Nonparametric confidence limits for survival probabilities and median survival time. Cancer Treat Rep. 1982;66:3742. 9. Efron B. Censored data and the bootstrap. JASA 1981;76:312–319.

Vol. 20 No. 2 August 2000

10. Mantel N, Haenszel W. Statistical aspects of the analysis of data from retrospective studies of disease. J Natl Cancer Inst 1959;22:71948. 11. Peto R, Pike MC, Armitage P, Breslow NE, Cox DR, Howard SV, et al. Design and analysis of randomized clinical trials requiring prolonged observation of each patient. Br J Cancer 1977;35:1–39. 12. Altman DG. Practical statistics for medical research. London: Chapman and Hall, 1991 13. Carmel R, Eisenberg L. Serum vitamin B12 and transcobalamin abnormalities in patients with cancer. Cancer 1977;40:1348–1353. 14. Beck WS. Metabolic aspects of vitamin B12 and folic acid. In: William JW, Beutler E, Erslew AJ, Lichtman MA, eds. Hematology. New York: McGraw Hill, 1983:311–331. 15. Hall CA. Les protéines de transport de cobalamines. In: Zittoun J, Cooper BA, eds. Folates and Cobolamines. Progrès en hématologie 9. Paris: Doin, 1987:51–66. 16. Carmel R, Hollander D. Extreme elevation of transcobalamin II levels in multiple myeloma and other disorders. Blood 1978;51:1057–1063. 17. Carmel R. Extreme elevation of serum transcobalamin I in patients with metastatic cancer. N Engl J Med 1975;292:282–284. 18. Hall CA, Wanko M. Increased transcobalamin I in a leukemoid reaction. J Lab Clin Med 1971;78: 298–301. 19. Neale G. B12 binding proteins. Gut 1990;Jan 31: 59–63. 20. Kanazawa S, Hervert V. Total corrinoid, cobalamin (vitamin B12) and cobalamin analogue levels may be normal in serum despite cobalamin in liver depletion in patients with alcoholism. Lab Invest 1985;53:108–110. 21. Djalali M, Champigneulle B, Gueant JL, el Kholty S, Gerard P, Nicolas JP. Increased serum corrinoids correlates with disease severity and IgA levels in alcoholic cirrhosis. Digestion 1998;41:215–222. 22. Grindularis KA, Calverley MK, Cox C. Rheumatoid arthritis: Is serum vitamin B12 high in active disease? J Rheumatol 1984;11:211–212. 23. Vreugdenhil G, Lindemans J, van Eijk HG, Swaak AJ. Elevated serum transcobalamin levels in anaemia of rheumatoid arthritis: correlation with disease activity but not with serum tumor necrosis factor alpha and interleukin 6. J Intern Med 1992; 231:547–550. 24. Arnalich F, Zamorano AF, Benito-Urbina S, Gijon-Banos J, DeMiguel E, et al. Increased apotranscobalamin II levels in rheumatoid arthritis. Br J Rheumatol 1990:171–173. 25. Arnalich F, Zamorano AF, Martinez-Hernandez P, Pena JM, Barbado FJ, Vazquez JJ. Additional value of serum unsaturated vitamin B12 proteins in multiple myeloma. J Med 1990;21:277–286.

Vol. 20 No. 2 August 2000

Elevated Vitamin B12 Predicts Mortality

26. Ogawa K, Shima N, Ohshio G, Kudo H, Nakashima Y, et al. Distribution of vitamin B12 R-binder in lung tumors. Implications for cell differentiation. Pathol Res Pract 1989;184:234–241. 27. Wakatsuki Y, Inada M, Kudo H, Oshio G, Masuda T, et al. Immunological characterisation and clinical implication of cobalamin binding protein in human gastric cancer. Cancer Res 1989;49:3122–3128. 28. Kudo H, Ohshio G, Ogawa K, et al. Distribution of vitamin B12 R-binder in carcinomas of the digestive tract. J Clin Pathol 1988;41:320–323. 29. Mazlam MZ, Hodgdson HJF. Why measure C-reactive protein? Gut 1994;35:5–7. 30. Stamatidis AP, Tourmandinou MS, Vyssoulis GP, Manouras AJ, Apostolidis NS. Value of serum acute-phase reactant proteins and carcinoembryonic antigen in the preoperative staging of colorectal cancer. Cancer 1990;65:2055–2057. 31. Merlini G, Perfetti V, Gobbi PG, Quaglini S, Franciotta DM et al. Acute phase proteins and prognosis in multiple myeloma. Br J Haematol 1993;83: 595–601. 32. Tienhaara A, Pulkki K, Mattila K, Irjala K, Pelliniemi TT. Serum immunoreactive interleukin-6 and C-reactive protein levels in patients with multiple myeloma at diagnosis. Br J Haematol 1994;86:391–393.

103

33. Baumann H, Gauldie J. The acute phase response. Immunol Today 1994;15:74–80. 34. Bruyère A, Delmi M, Pochon N, et al. Les paramètres biologiques nutritionnels peuvent nous aider à dèfinir une stratégie thérapeutique en soins palliatifs. Méd et Hyg 1991;49:3144–3152. 35. Romagnoli-Juillard A. Malnutrition en gériatrie: incidence et prognostic. Thèse de médecine. Genève 1989. No. 9036. 36. Maltoni M, Nanni O, Derni S, Innocenti MP, Fabbri L, et al. Clinical prediction of survival is more accurate than the Karnofsky perfomance status in estimating life span of terminally ill cancer patients. Eur J Cancer 1994;3OA:764–766. 37. Lovey M. Ethique et fin de vie: vécu d’un infirmier référent et du médecin assistant confrontés á la période terminale d’un patient dans un service hospitalier universitaire. Thèse de médecine No. 9732, Université de Genève, 1996. 38. Winslow GR. From loyalty to advocacy: a new metaphor for nursing. Hastings Cent Rep 1984;14:32–35. 39. Lubitz JD, Riley GF. Trends in medicare payments in the last year of life. N Engl J Med 1993;328: 1092–1096. 40. Felder S. Costs of dying: alternatives to rationing. Health Policy 1997;39:167–176.