Comments on News and Primary Papers
Primary Papers: Quality control for microarray analysis of human brain samples: The impact of postmortem factors, RNA characteristics, and histopathology.Comment by: Marquis Vawter
Submitted 21 September 2007
Posted 21 September 2007
This article by senior author Dr. Christine Miller and colleagues is one of several recent articles that demonstrate the rigorous assessments required to derive useful knowledge in high-throughput gene expression experiments. The well-known adage that “garbage in = garbage out” was not surprisingly demonstrated in this careful work by first author S. Weis et al.
The authors demonstrated many additional points. One salient point was the relationship of cerebellar acidosis and neuronal atrophy, i.e., as pH decreases there was more cerebellar granular cell layer necrosis. This finding is consistent with previous reports such as Dr. Paul Harrison’s 1995 work: “The relative importance of premortem acidosis and postmortem interval for human brain gene expression studies: selective mRNA vulnerability and comparison with their encoded proteins” (Harrison et al., 1995) and even earlier studies of glutamic acid decarboxylase activity alterations in brain tissue from patients with Huntington's disease in the 1970s-1980s (Spokes et al., 1979; Spokes, 1979; Spokes, 1980.
Second, perhaps a finer point that may have the field rethinking some conclusions is that pH appears to be lower among various patient groups by 0.08 pH units. The number of subjects used for the pH measures was perhaps the largest yet reported in postmortem human brain for schizophrenia (n = 120), controls (n = 116), and bipolar disorder (n = 113). Interestingly, comparing control subjects with obvious histopathological abnormalities—(cerebellar granular cell layer necrosis (CGLN)—to control subjects without CGLN showed a decreased pH of 0.4. However, when comparing CGLN+ to CGLN- schizophrenia groups, the pH was reduced only 0.04 pH units, which was not significant. This diagnosis group x CGLN interaction for pH is one example of how the field is grappling with issues of how to control for these strong associations of postmortem gene expression overall. It would be useful to run some linear regression models to predict pH using the arsenal of data available at the Stanley Foundation-557 collection to determine the strongest predictors of pH. Beyond the usual predictors of age, PMI, agonal factors, and tissue necrosis, it would be interesting to see if the lifetime fluphenazine daily equivalents, smoking, and other variables were strongly predictive.
If the decreased pH between psychiatric groups and controls could be replicated in other vulnerable regions such as the hippocampus, and were found in rapid death cases, then the pathophysiology of lower pH in bipolar disorder and schizophrenia might be considered to be a general feature, as has been previously suggested (Konradi et al., 2004; Altar et al., 2005; Prabakaran et al., 2004; Iwamoto et al., 2005, and others). Further, this would suggest that animal and cellular models that mimic lower antemortem pH and produce abnormalities in gene expression pathways similar to bipolar and schizophrenia could be useful.
However, there are some caveats to these findings. Another recent article also demonstrates the care and rigorous methodology required to produce useful microarray data. Atz and colleagues also addressed critical variables that might affect gene expression in two brain regions using measures from over 91 subjects (Atz et al., 2007). Four broad groups of quality indicators in gene expression profiling studies (clinical, tissue, RNA, and microarray quality) were identified. These quality control indicators were significantly correlated; however, no single quality variable accounted for the total variance in microarray gene expression. On the contrary, this theme suggests that pH, agonal state, along with other gene expression covariates such as RNA quality, gender, and aging, must be strictly controlled. A researcher, when evaluating individual candidate genes, can use these quality parameters (clinical, tissue, RNA, and microarray quality) in post hoc analysis to help strengthen the relevance to psychiatric disorders. Further, the Atz et al. data showed that agonal factors and lower pH correlated with decreased integrity of extracted RNA in two brain regions. This relationship for pH was not significant in Weis et al. for microarray quality except for one measure of GAPDH 3'/5' ratio, and agonal factors were not compared in Weis et al.. However, in Atz et al. three parameters (agonal factors, pH, and RNA quality) also modulated the significance of alterations in mitochondrial-related genes. The average F-ratio summaries across all transcripts showed that RNA degradation from the AffyRNAdeg program accounted for higher variation than all other quality factors. The AffyRNAdeg program only infers RNA degradation from post hoc microarray analysis.
Taken together, the Atz et al. and Weis et al. findings, as well as a 2006 study by Lipska and colleagues (Lipska et al., 2006), are rigorous methodological studies of human postmortem gene expression, and all suggest that quality parameters, including RNA integrity, are related to differences in gene expression profiles in postmortem brain. However, some distinction in these recent reports shows that agonal factors and pH were more prominent in some researchers' gene expression findings (Vawter et al., 2006; Iwamoto et al., 2005; Mexal et al., 2006). Clearly Atz et al. show a pH relationship to microarray quality, while Weis et al., 2007, do not find a relationship. If Weis et al. reported clustering or principal components analysis, this might indicate beyond RNA quality what other factors are impacting the results. Since pH was strongly related to CGLN, then it seems that pH would also affect the gene expression profile in Weis et al.
In a third recent article, Arion and colleagues confronted the issue of candidate gene expression in postmortem brain by using a custom microarray with quintuplicate measurements of the same gene and running the experiment on three separate arrays hybridized from separate biological replicates (Arion et al., 2007). The postmortem pH measurements were group-matched and pair-matched for each sample closely, and all subjects had rapid deaths and were not hospitalized prior to death. The use of rigorous experimental design and statistical techniques (paired t-test and group average testing) led to 67 genes that survived four rounds of testing, which yielded reliable fold changes, and with reduced false positive results. The authors comment that false negatives might not be adequately controlled for in the design. Arion and colleagues indicate that many other interesting genes, which would pass less stringent testing, are probably among their results and offer these lists to qualified researchers. The QPCR confirmation was indeed very robust and the paper proffered candidate immune related genes. These findings mesh well with different environmental and genetic theories of schizophrenia pathophysiology. In my own view, this paper addresses by experimental design the strong effects of agonal status and pH, gender, age, RNA quality, and microarray quality, on gene expression. Nevertheless, it would be of interest to see if the findings were pH independent—if the effect were fairly constant across the pH range studied in the Arion et al. report—which would rule out uncontrolled agonal factors in the Arion study, especially at the lower pH range. Since these subjects are exquisitely matched, then a smoking gun is nearby, and these might be part of the premortem pathophysiology of schizophrenia.
This series of papers draws attention to the strict methodological rigor that will be required at multiple levels of investigation for the next generation of microarray studies to lead the field of gene expression toward replicable and meaningful pathways of gene expression underlying neuropsychiatric disorders. It seems that the field has been able to address the overarching strong covariates, and now several viable methods shown in these recent papers and by many other groups not mentioned will lead to identification of genes associated with psychiatric diagnosis.
Harrison PJ, Heath PR, Eastwood SL, Burnet PW, McDonald B, Pearson RC. The relative importance of premortem acidosis and postmortem interval for human brain gene expression studies: selective mRNA vulnerability and comparison with their encoded proteins. Neurosci Lett. 1995 Nov 24;200(3):151-4. Abstract
Spokes EG, Garrett NJ, Iversen LL. Differential effects of agonal status on measurements of GABA and glutamate decarboxylase in human post-mortem brain tissue from control and Huntington's chorea subjects. J Neurochem. 1979 Sep 1;33(3):773-8. Abstract
Spokes EG. GABA in Huntington's chorea, Parkinsonism and schizophrenia. Adv Exp Med Biol. 1979 Jan 1;123():461-73. Abstract
Spokes EG. Neurochemical alterations in Huntington's chorea: a study of post-mortem brain tissue. Brain. 1980 Mar 1;103(1):179-210. Abstract
Konradi C, Eaton M, MacDonald ML, Walsh J, Benes FM, Heckers S. Molecular evidence for mitochondrial dysfunction in bipolar disorder. Arch Gen Psychiatry. 2004 Mar 1;61(3):300-8. Abstract
Altar CA, Jurata LW, Charles V, Lemire A, Liu P, Bukhman Y, Young TA, Bullard J, Yokoe H, Webster MJ, Knable MB, Brockman JA. Deficient hippocampal neuron expression of proteasome, ubiquitin, and mitochondrial genes in multiple schizophrenia cohorts. Biol Psychiatry. 2005 Jul 15;58(2):85-96. Abstract
Prabakaran S, Swatton JE, Ryan MM, Huffaker SJ, Huang JT, Griffin JL, Wayland M, Freeman T, Dudbridge F, Lilley KS, Karp NA, Hester S, Tkachev D, Mimmack ML, Yolken RH, Webster MJ, Torrey EF, Bahn S. Mitochondrial dysfunction in schizophrenia: evidence for compromised brain metabolism and oxidative stress. Mol Psychiatry. 2004 Jul 1;9(7):684-97, 643. Abstract
Iwamoto K, Bundo M, Kato T. Altered expression of mitochondria-related genes in postmortem brains of patients with bipolar disorder or schizophrenia, as revealed by large-scale DNA microarray analysis. Hum Mol Genet. 2005 Jan 15;14(2):241-53. Abstract
Atz M, Walsh D, Cartagena P, Li J, Evans S, Choudary P, Overman K, Stein R, Tomita H, Potkin S, Myers R, Watson SJ, Jones EG, Akil H, Bunney WE, Vawter MP, . Methodological considerations for gene expression profiling of human brain. J Neurosci Methods. 2007 Jul 30;163(2):295-309. Abstract
Lipska BK, Deep-Soboslay A, Weickert CS, Hyde TM, Martin CE, Herman MM, Kleinman JE. Critical factors in gene expression in postmortem human brain: Focus on studies in schizophrenia. Biol Psychiatry. 2006 Sep 15;60(6):650-8. Abstract
Vawter MP, Tomita H, Meng F, Bolstad B, Li J, Evans S, Choudary P, Atz M, Shao L, Neal C, Walsh DM, Burmeister M, Speed T, Myers R, Jones EG, Watson SJ, Akil H, Bunney WE. Mitochondrial-related gene expression changes are sensitive to agonal-pH state: implications for brain disorders. Mol Psychiatry. 2006 Jul 1;11(7):615, 663-79. Abstract
Mexal S, Berger R, Adams CE, Ross RG, Freedman R, Leonard S. Brain pH has a significant impact on human postmortem hippocampal gene expression profiles. Brain Res. 2006 Aug 23;1106(1):1-11. Abstract
Arion D, Unger T, Lewis DA, Levitt P, Mirnics K. Molecular evidence for increased expression of genes related to immune and chaperone function in the prefrontal cortex in schizophrenia. Biol Psychiatry. 2007 Oct 1;62(7):711-21. Abstract
View all comments by Marquis Vawter