This is the first study in which the prognostic value of oxidative stress markers and antioxidant enzymes in DLBCL patients treated by means of modern immunochemotherapy has been assessed. We discovered that low immunoexpression of GCL, Trx and nitrotyrosine was associated with favorable progression-free survival and Trx was also significant as regards disease-specific survival. Intensive Trx immunostaining also revealed a high-risk group when only non-GC phenotype patients were included. Staining of GCL was almost significant as regards prediction of poor disease-specific survival. Our results suggest that excessive antioxidant and high nitrotyrosine expression are associated with adverse prognosis.
The prognostic role of oxidative stress in lymphomas has been studied previously, with conflicting results. A study by Tome et al. (2005)  concerned analysis of the levels of antioxidant enzymes and redox-associated proteins. Their results suggested that the group with the worst prognosis were those with decreased expression of antioxidant enzymes, including catalase, glutathione peroxidase and MnSOD, and increased expression of Trx. In a study by Andreadis et al. (2007) , the expression of genes in the glutathione family was examined. Their data suggested that overexpression of these genes correlated with worse prognosis. They used the redox signature score created by Tome et al.
 and found that their results were exactly the opposite, i.e. overexpression of antioxidants and redox-state proteins correlated with poor prognosis. However, in neither of these studies had the patients received modern immunochemotherapy. Because prognostic markers are highly associated with the therapeutic modality used, the results cannot be extrapolated to patient populations undergoing modern treatments. We have previously carried out thorough evaluation of the expression of oxidative stress markers in different forms of lymphoma. In a recent study  we found that high levels of GCL and nitrotyrosine, and also low levels of MnSOD correlated with poor prognosis, but the number of patients with DLBCL in that study was only 18. The aim of the present study was to assess the relationships between antioxidant enzymes, oxidative stress markers and prognosis in cases of diffuse large B-cell lymphoma treated in the rituximab era.
At first glance the results reported by Andreadis et al. and Tome et al. seem to be contradictory, but when evaluating the data according to the expression of individual enzymes, the differences are not so striking. Comparing the results reported by Tome et al. and those in the current study, most are contradictory, as their data indicated that antioxidant function would be a cellular protector. Both our study and theirs showed that strong intensity of Trx was associated with worse prognosis. In the study by Andreadis et al. the authors came to the same conclusion as ourselves, i.e. that the GSH system is related to poor prognosis and excessive antioxidant expression worsens the prognosis. Because the biological functions are complex and antioxidant enzymes may have both positive and negative effects on malignant cell growth, the expression of these markers should probably be analyzed individually and also possibly disease-specifically. Redox state-regulating enzymes may have prognostic value regardless of treatment, since our results are largely in line with those in two previous studies where different treatments were applied. The expression of 8-OHdG, the most commonly used marker of oxidative DNA damage, is a significant prognostic factor in various solid malignancies, but in DLBCL it does not seem to have a prognostic role. Nevertheless, high cytoplasmic 8-OHdG expression, which reflects ROS-derived mitochondrial DNA adduct formation, was associated with features of aggressive disease such as high IPI class and extranodal involvement. Expression of 8-OHdG was associated with high Trx and GCL expression and this suggests therefore that these enzymes might be induced under heavy oxidative stress in DLBCL cells.
Rituximab has revolutionized DLBCL treatment and patient prognosis. However, some patients still succumb to their lymphomas. IPI scoring is at the moment the only clinically relevant prognostic scoring system. Results regarding the prognostic value of gene expression profiling-based grouping are contradictory [25, 26]. Unfortunately, IPI scoring can no longer identify a patient population with a 5-year prognosis of less than 50%. Therefore, reliable biological prognostic factors and especially factors describing basic biological phenomena which might serve as predictors of treatment results are needed. Our results indicate that oxidative stress markers have correlations with prognosis in cases of DLBCL. Preclinical data indicate that GSH induces resistance against key lymphoma chemotherapeutic agents. In vitro, drugs such as indomethacin can counteract this adverse effect by reducing GSH levels . This implies that drugs that suppress GSH might be therapeutically beneficial in lymphomas with high GSH expression. This may also have relevance to the Trx system . Trx is an activator of nuclear factor κB (NF-κB), which inhibits apoptosis and increases proliferation . Non-GC-type DLBCL is characterized by constitutional activation of NF-κB. It has been suggested that high levels of NF-κB may be a sign of adverse prognosis [30, 31]. Our observation of strong Trx intensity and poor prognosis in non-GC phenotype patients might be explained by activation of NF-κB and its anti-apoptotic properties through Trx overexpression.
Resistance to doxorubicin  and vincristine  has been linked to increased amounts of intracellular GSH. Cyclophosphamide treatment depletes GSH and increases the amount of oxidative stress in human ovarian granulosa cells . Therefore, it is plausible that an increased amount of GSH limits the effectiveness of cyclophosphamide and might be connected to lymphoma chemoresistance. Our results may partly be explained by activation of GCL, which increases the amount of GSH and may therefore bring about resistance to chemotherapy. Doxorubicin sensitivity has also been linked to an excessive amount of Trx .