Enhanced cyclooxygenase-2 (COX-2) expression has been reported for many human cancers and is often considered a feature of aggressive phenotype and/or poor prognostic factor. Jang et al.1 assessed the COX-2 expression in 414 cases of colorectal cancers and concluded that “COX-2 overexpression was significantly associated with favorable clinicopathologic phenotype and an indicator of better survival … .”
We suggest that the authors’ conclusion was due rather to improper interpretation of the results of their immunohistochemical study than to biological features of the cancers investigated. The authors used two approaches to evaluate the COX-2 expression: staining intensity (I = 0–3) and staining extent (SE = 0–4 depending on percentage fraction of the stained cells). The two parameters are usually analyzed separately (sometimes their ratio and/or product are also calculated). Surprisingly, Jang et al.1 decided to use a sum of I and SE for evaluation of the COX-2 expression levels classifying the cases with I + SE of less than 4 as negative and those with I + SE of 4 or more as overexpressing the protein.
It is generally not acceptable to calculate whatever parameter as a sum of quantities measured in different units. Moreover, it is easy to demonstrate that the (I + SE) parameter yields confusing results.
According to interpretation rules used by Jang et al.,1 the I + SE parameter of 4 (lowest value found for “overexpressing” cancers) can be obtained for several combinations given in Table 1. Clearly, cases 1 and 2 are very different in terms of biological features related to the COX-2 expression and progress of the disease had been probably different in those two cases. Also, taking into account subjective assessment of I and unavoidable errors in the SE determination, it is difficult to accept that cases of undistinguishable staining characteristics would be classified as either non-expressing or overexpressing COX-2 (compare, for instance, the situations of I + SE = 3 for combinations like 3 = 2 (moderate intensity) + 1 (25%) or 3 = 1 (weak intensity) + 2 (50%) with “overexpressing” cases of I + SE = 4 shown in rows 3 and 1 of Table 1, respectively).
|Case no.||I + SE = 4 “overexpression”||I||SE|
|1||4 = 1 + 3||Weak – 1||51%– 3|
|2||4 = 3 + 1||Strong – 3||1%– 1|
|3||4 = 2 + 2||Moderate – 2||26%– 2|
|4||4 = 2 + 2||Moderate – 2||50%– 2|
We strongly believe that the immunohistochemical results reported by Jang et al.1 need reinterpretation and should be analyzed and presented in a commonly acceptable manner to verify the authors’ conclusions concerning association of COX-2 overexpression with favorable prognosis.