This is, of course, a complex question and the prior answers have only added levels of complexity.  Tumoro neuropathology is inevitably and properly going the way of hematopathology--molecular genetic (and epigenetic) information adds precision to prognostication and to choices of therapy and so becomes an essential part of the diagnosis; pretending otherwise (and staying with H&E alone) is just putting one's head in the sand.

That said, the molecular genetic data need to be developed with objectivity and with correlation to histological and immunohistological data.  In my opinion the conclusion that 1p/19q co-deletion is "diagnostic" of oligodendroglioma was jumped at too rapidly, and did not account for (1) classical oligodendroglioma by histology which are not co-deleted, in adults (rare but definitely occurs), (2) pediatric oligodendrogliomas which are usually not co-deleted; (3) the variably demonstrable presence of neuronal differentiation using ultrastructural or immunohistochemical markers of such differentiation in so-called oligodendrogliomas, with and without 1p/19q deletions; (4) the rare demonstration of 1p/19q co-deletion in tumors not histologically compatible with a diagnosis of oligodendroglioma.  When one considers the enormous spatial heterogeneity in any large resection specimen of diffuse gliomas, whether looking at histology, at immunostains as markers of protein expression, or at molecular genetic criteria, it becomes clear that looking at the histological diagnosis of OA and then taking one sample for genetic analysis and rejecting the mixed diagnosis because there is either 1p/19q co-deletion in that sample or because there is p53 mutation and ATRX mutation in that sample is inappropriate.

Histological appearances presumably result from an interaction with tumor genetic characteristics and environmental influences.  We understand little about the genes suppressed or activated in 1p/19q co-deleted gliomas, but the available studies point to abnormal regulation of genes controlling neuronal differentiation.  Taking the cancer stem cell hypothesis at face value here, OA is the resujlt of differential activation or suppression of neuronal differentiation factors in different cells or different parts of these tumors, under the control or influence of local environmental factors about which we know almost nothing.  So it is premature to think that the diagnosis of OA can be legislated out of existence.

My own diagnostic practices have evolved from how I understand the literature such that I now more frequently make a diagnosis of "glioblastoma with oligodendroglioma" to recognize a tumor which has histological components of oligodendroglioma or neurocytoma but which also has necrosis with pseudopalisades and florid vascular hyperplasia.  Similarly, I find I see more cases of "glioblastoma with PNET" in adults than I used to recognize.  By using panels of immunohistochemical markers I have for a long time recognized that neuronal differentiation is more common in all of these tumors than standard texts, or the WHO, concede.  Whether this is of clinical importance is not wholly clear, although some studies from the Paris group have shown that glioblastomas with neurofilament protein immunopositivity have a better prognosis compared to those that do not.

The important data from the past few years suggests that the histological diagnosis may be of less therapeutic ("predictive") and prognostic importance as compared to the genetic data; thus once one identifies a diffuse glioma, knowing that it has 1p/19q co-deletion, or IDH mutation, or ATRX mutation, appears to be more important than whether one calls it from histology astrocytoma, oligodendroglioma, parenchymal neurocytic tumor ("extraventricular neurocytoma"), or mixed glioma (OA).  For practical, clinical purposes, this information assumes paramount importance now, until we better understand the factors regulating glioma cell differentiation and whether that understanding will lead to better targeted therapies.

One last note--the WHO won't be coming out with a wholly new CNS tumor classification for some time.  The "interim" update from the Haarlem meeting was "allowed", as I understand what was said by leaders of that meeting, because the WHO was pressured by the neuro-oncology community to recognize the significant changes that have occurred since the 2008 publication of the last WHO "blue book", and this has happened also in the hematopathology world (in fact, more so).  However the WHO bureaucracy is apparently wedded to finishing the current series of blue books across all organ systems and cancer types before allowing work to begin on the next series, so a wholly new WHO CNS tumor book can't be expected until late in this decade, if then.  Don't hold your breath.

1. In my opinion neuropathologist should retain the role and responsibility of synthesis, avoiding a fragmented diagnosis.

2. Some Authors state that bio-molecular assessment offers far more potential for objectivity. This may be true if bio-molecular investigations are strictly associated to an accurate histological evaluation, able to recognize the representative areas, for example in an oligoastrocytoma displaying a biphasic pattern.

3. World Health Organization Classification represents an international reference. Until today, applying WHO criteria, everyone could diagnose a glioblastoma or an anaplastic astrocytoma with H&E, everywhere in the world, also in low- or middle income countries. And tomorrow?

My two cents (for what it's worth)...

It will be really interesting to see how the upcoming WHO changes the classifications. I have also seen a great deal of inter-observer variability in the diagnosis of OAs. And then of course, the complications and differences in how fellow neuropathologists upgrade one (i.e.:  can an AOA, WHO grade III become a GBM?)

I do think that the true nature of these lesions lies in their molecular signature. In a perfect world where everyone could have standardized molecular testing on a regular basis, a combination of the histopathologic and molecular diagnosis would allow for the most complete diagnosis. But I realize that many hospitals and institutions do not have access to molecular testing. In addition, even institutions that perform "molecular" testing may be looking at different output parameters (i.e.: RNA, copy number, protein expression, etc).

Perhaps the new WHO will allow for multiple levels of diagnosis to incorporate the molecular component in addition to the histopathologic diagnosis while allowing for the later alone. But something I do feel strongly about is that the neuropathologist should be involved in the molecular testing. We need to have the knowledge and ability to direct the testing and be at least partially involved in the interpretation. At our institution, any additional molecular testing whether it be in-house or send-out is used directly by our neuro-oncologists to determine treatment strategies. I know that some are hesitant to move toward molecular diagnostics, but in the end, the point of all of our jobs is to get the best outcome for our patients.

This is, of course, a complex question and the prior answers have only added levels of complexity.  Tumoro neuropathology is inevitably and properly going the way of hematopathology--molecular genetic (and epigenetic) information adds precision to prognostication and to choices of therapy and so becomes an essential part of the diagnosis; pretending otherwise (and staying with H&E alone) is just putting one's head in the sand.

That said, the molecular genetic data need to be developed with objectivity and with correlation to histological and immunohistological data.  In my opinion the conclusion that 1p/19q co-deletion is "diagnostic" of oligodendroglioma was jumped at too rapidly, and did not account for (1) classical oligodendroglioma by histology which are not co-deleted, in adults (rare but definitely occurs), (2) pediatric oligodendrogliomas which are usually not co-deleted; (3) the variably demonstrable presence of neuronal differentiation using ultrastructural or immunohistochemical markers of such differentiation in so-called oligodendrogliomas, with and without 1p/19q deletions; (4) the rare demonstration of 1p/19q co-deletion in tumors not histologically compatible with a diagnosis of oligodendroglioma.  When one considers the enormous spatial heterogeneity in any large resection specimen of diffuse gliomas, whether looking at histology, at immunostains as markers of protein expression, or at molecular genetic criteria, it becomes clear that looking at the histological diagnosis of OA and then taking one sample for genetic analysis and rejecting the mixed diagnosis because there is either 1p/19q co-deletion in that sample or because there is p53 mutation and ATRX mutation in that sample is inappropriate.

Histological appearances presumably result from an interaction with tumor genetic characteristics and environmental influences.  We understand little about the genes suppressed or activated in 1p/19q co-deleted gliomas, but the available studies point to abnormal regulation of genes controlling neuronal differentiation.  Taking the cancer stem cell hypothesis at face value here, OA is the resujlt of differential activation or suppression of neuronal differentiation factors in different cells or different parts of these tumors, under the control or influence of local environmental factors about which we know almost nothing.  So it is premature to think that the diagnosis of OA can be legislated out of existence.

My own diagnostic practices have evolved from how I understand the literature such that I now more frequently make a diagnosis of "glioblastoma with oligodendroglioma" to recognize a tumor which has histological components of oligodendroglioma or neurocytoma but which also has necrosis with pseudopalisades and florid vascular hyperplasia.  Similarly, I find I see more cases of "glioblastoma with PNET" in adults than I used to recognize.  By using panels of immunohistochemical markers I have for a long time recognized that neuronal differentiation is more common in all of these tumors than standard texts, or the WHO, concede.  Whether this is of clinical importance is not wholly clear, although some studies from the Paris group have shown that glioblastomas with neurofilament protein immunopositivity have a better prognosis compared to those that do not.

The important data from the past few years suggests that the histological diagnosis may be of less therapeutic ("predictive") and prognostic importance as compared to the genetic data; thus once one identifies a diffuse glioma, knowing that it has 1p/19q co-deletion, or IDH mutation, or ATRX mutation, appears to be more important than whether one calls it from histology astrocytoma, oligodendroglioma, parenchymal neurocytic tumor ("extraventricular neurocytoma"), or mixed glioma (OA).  For practical, clinical purposes, this information assumes paramount importance now, until we better understand the factors regulating glioma cell differentiation and whether that understanding will lead to better targeted therapies.

One last note--the WHO won't be coming out with a wholly new CNS tumor classification for some time.  The "interim" update from the Haarlem meeting was "allowed", as I understand what was said by leaders of that meeting, because the WHO was pressured by the neuro-oncology community to recognize the significant changes that have occurred since the 2008 publication of the last WHO "blue book", and this has happened also in the hematopathology world (in fact, more so).  However the WHO bureaucracy is apparently wedded to finishing the current series of blue books across all organ systems and cancer types before allowing work to begin on the next series, so a wholly new WHO CNS tumor book can't be expected until late in this decade, if then.  Don't hold your breath.

I managed to hold my breath just long enough. Well, it's only been ~17 months since I posted the question initially and 9 years since the last major edition release, but about a week ago the WHO published a revised version of the classification (http://apps.who.int/bookorders/anglais/detart1.jsp?codlan=1&codcol=70&codcch=24001) and a summary in Acta (http://link.springer.com/article/10.1007/s00401-016-1545-1). So oligoastrocytoma remains as a possible diagnosis but only as "NOS" in the absence of molecular testing, otherwise classified as oligodendroglioma, IDH-mutant and 1p/19q-codeleted, diffuse astrocytoma, IDH-mutant or their grade-III anaplastic counterparts.

Thanks for letting us know about the new publication...I had missed the notice of publication of the book, and I have now ordered my copy and will have my Department buy copies for our residents and our signout room.