We read with interest the recent report of Ishizuka and colleagues. They examined the prognostic value of markers of the systemic inflammatory response C-reactive protein (>10 mg/L) or C-reactive protein (>10 mg/L) and albumin (<35 g/L) combined (Glasgow Prognostic Score, GPS) in 93 patients with colorectal liver metastases. They reported that an elevated C-reactive protein but not the GPS predicted survival independent of conventional staging criteria and was superior to CEA and CA19-9 in predicting cancer specific mortality.
The authors speculate why C-reactive protein alone was superior to the GPS in predicting survival. They suggest two possibilities, (1) given the advanced stage disease in this cohort, mortality may be high irrespective of the GPS and (2), albumin was not a prognostic indicator and simply weakens the GPS. With reference to the latter, it has been recognised for some time that, in patients with different advanced cancers, hypoalbuminaemia rarely occurs (approximately 10%) in absence of an elevated C-reactive protein [1-4]. However, it was recognised that the prognostic value of hypoalbuminaemia in absence of an elevated C-reactive protein may be different. This was investigated by McMillan et al. [5] who in patients with primary operable colorectal cancer reported that of hypoalbuminaemia in absence of an elevated C-reactive protein (16 out of 54, 30%) was not independently associated with poor survival. Therefore, the GPS was modified (mGPS) such that patients with an elevated C-reactive protein and hypoalbuminaemia scored 2, an elevated C-reactive protein scored 1 and a normal C-reactive protein with any albumin scored 0. The mGPS has been used in subsequent publications from this group [6][7][8][9][10].
Therefore, in the present study of Ishizuka and colleagues, where approximately 65% (11 out of 17) patients with hypoalbuminaemia did not have an elevated C-reactive protein, the use of the GPS instead of the mGPS is likely to have had a profound detrimental effect on its prognostic value.