On 2014 Oct 31, Francisco Felix commented:

The many technical flaws in this article should be obvious to the educated reader, but I can summarize the most prominent ones for the sake of helping the eventual reader of scientific papers. 1. Diagnostic criteria: even though there is no clear cut criteria listed for DIPG, there is a common set of characteristics for defining it. As the acronym implies, it should be DIFFUSE, meaning no evident boundaries between the lesion and surrounding brain tissue. Also, it has to be INTRINSIC, meaning it should be completely embedded in pons, with no main exophytic or extra-axial portion. Other commonly recognized features include little or no contrast enhancement of the lesion and location on the ventral aspect of pons. For my astonishment, figure 2 depicts a mesencephalic focal tumor with cystic areas, typical of a low grade glioma with good prognosis, whereas figure 3 shows an upper pons, lower mesencephalon focal, highly enhancing lesion. 2. Accrual time: why it was so long? What happened to the patients diagnosed between 1994 and 2003? This is not a trivial question. We know that DIPG comprises 15-20% of all pediatric brain tumors, so there should be a regular registration of new patients in the trial. Around 3-5% of DIPG patients show longer survival, and duration of symptoms before diagnosis and patient age are strong factors influencing survival. This patient series seems to be enriched in older patients and longer pre-diagnostic symptom duration, both factors associated with better survival. This strongly indicates a bias, and one could imagine if there was some intentional selection of patients. 3. Evaluation: once there is irregular contrast enhancement by DIPG lesions, it is simply wrong to evaluate response by measuring it. Period. Hence, no result pertaining response evaluation should be trusted in this paper. 4. Prior treatment: patients with more than one previous chemotherapy treatment should raise suspicion, because tipically DIPG patients do not survive beyond the first progression. So, a second and even a third progression seems highly unlikely. 5. Survival data: there is no median survival time reported, and the graph difficults its visual inspection. However, it probably lies before 6 months, once survival was less than 30% at one year. The lack of confidence intervals make it hard to define if this survival time reflects reality. The best estimate on literature (upper bound) is around 5 months in patients that have done radiotherapy after progression. So, where is the point? 6. Title: there is no such thing as a 'recurrent DIPG', once there are no complete responses to treatment for this disease. It should refer to 'progressive DIPG'. However, this is such a small detail when compared to the many flaws of this article that it is almost worthless to mention.

On 2014 Oct 31, Francisco Felix commented:

The many technical flaws in this article should be obvious to the educated reader, but I can summarize the most prominent ones for the sake of helping the eventual reader of scientific papers. 1. Diagnostic criteria: even though there is no clear cut criteria listed for DIPG, there is a common set of characteristics for defining it. As the acronym implies, it should be DIFFUSE, meaning no evident boundaries between the lesion and surrounding brain tissue. Also, it has to be INTRINSIC, meaning it should be completely embedded in pons, with no main exophytic or extra-axial portion. Other commonly recognized features include little or no contrast enhancement of the lesion and location on the ventral aspect of pons. For my astonishment, figure 2 depicts a mesencephalic focal tumor with cystic areas, typical of a low grade glioma with good prognosis, whereas figure 3 shows an upper pons, lower mesencephalon focal, highly enhancing lesion. 2. Accrual time: why it was so long? What happened to the patients diagnosed between 1994 and 2003? This is not a trivial question. We know that DIPG comprises 15-20% of all pediatric brain tumors, so there should be a regular registration of new patients in the trial. Around 3-5% of DIPG patients show longer survival, and duration of symptoms before diagnosis and patient age are strong factors influencing survival. This patient series seems to be enriched in older patients and longer pre-diagnostic symptom duration, both factors associated with better survival. This strongly indicates a bias, and one could imagine if there was some intentional selection of patients. 3. Evaluation: once there is irregular contrast enhancement by DIPG lesions, it is simply wrong to evaluate response by measuring it. Period. Hence, no result pertaining response evaluation should be trusted in this paper. 4. Prior treatment: patients with more than one previous chemotherapy treatment should raise suspicion, because tipically DIPG patients do not survive beyond the first progression. So, a second and even a third progression seems highly unlikely. 5. Survival data: there is no median survival time reported, and the graph difficults its visual inspection. However, it probably lies before 6 months, once survival was less than 30% at one year. The lack of confidence intervals make it hard to define if this survival time reflects reality. The best estimate on literature (upper bound) is around 5 months in patients that have done radiotherapy after progression. So, where is the point? 6. Title: there is no such thing as a 'recurrent DIPG', once there are no complete responses to treatment for this disease. It should refer to 'progressive DIPG'. However, this is such a small detail when compared to the many flaws of this article that it is almost worthless to mention.