If you look at the conclusion by the authors of this excellent piece of research, it seems pretty clear that neurocognitive variables add nothing to the predictive value of the clinical variables alone, which is what we have found in our research, too. So, sadly, predictive power of the neurocognitive variables is no help. There has been a lot of hype regarding cognition in schizophrenia, but little useful progress, and this excellent paper illustrates that, in my view.

View all comments by Patrick McGorry

Reply to comment by Patrick McGorry
In response to Professor McGorry's comments, it is hard to imagine that neurocognitive measures really “add nothing” to the prediction of psychosis. The Seidman paper focuses primarily on neuropsychological measures that were collected in each of several research projects. They tended to be standard tests that are clear measures of general neuropsychological function known to be severely impaired in schizophrenia. However, the breakdown of cognitive function that occurs prior to the onset of psychosis may require more specific cognitive indices that best reflect how an individual maintains a sense of reality.

How the human brain staves off hallucinations and delusions is likely not a cognitive function that is easily measured with standard neuropsychology. Recent models describe how perception depends upon learning-dependent prediction, and how impairments in these cognitive processes may lead to chronic misperception of reality and subsequent hallucinations and delusions (Keefe et al., 2009; Corlett et al., 2010; Krishnan et al., in press; Fletcher and Frith, 2009). Research on specific cognitive mechanisms that may underlie the onset of psychosis is needed.

References:

Keefe RSE, Kraus MS. Measuring memory-prediction errors and their consequences in youth at risk for schizophrenia. Ann Acad Med Singapore. 2009;38:414-6. Abstract

Corlett PR, Taylor JR, Wang XJ, Fletcher PC, Krystal JH. Toward a neurobiology of delusions. Prog Neurobiol. 2010 June 15. Abstract

Krishnan RR, Fivaz M, Kraus M, Keefe RSE. Hierarchical temporal processing deficit model of schizophrenia. Mol Psych. In press.

Fletcher PC, Frith CD. Perceiving is believing: A Bayesian approach to explaining the positive symptoms of schizophrenia. Nat Rev Neurosci. 2009;10:48-58. Abstract

View all comments by Richard Keefe

  I recommend this paper

I agree with the previous two commentators that this is an excellent paper. An interesting finding was that neuropsychological difficulties were present in clinical high-risk (CHR) individuals who later developed psychosis, but they were not as marked as impairments found in people with first-episode schizophrenia. The authors commented that this phenomenon might be evidence of cognitive decline occurring during the prodromal phase. This could indicate that a “process” is occurring, suggesting a discontinuity between pre-psychotic and psychotic states, rather than just an accumulation of psychotic symptoms of increasing frequency and intensity. Wood et al. found evidence of cognitive decline in the peri-onset phase in ultra high-risk individuals, but this finding was not supported by Hawkins et al.

Related to the issue of whether there is any discontinuity between the prodrome and first-episode psychosis is the question of whether we have the threshold of “psychosis” right. As we have previously discussed (Yung et al., 2010), it may be possible, for example, that some people may cross the arbitrary threshold of “psychosis,” but then quickly return to a level of symptoms below the threshold. Examination of whether such “trivial” transitions exist, and their implications for long-term outcome can only be determined by longitudinal studies. Longitudinal neurocognitive assessment adds a further dimension to validating the “psychosis threshold.”

The other explanation for the finding of less severe neurocognitive impairments in the CHR group who became psychotic compared to the first episode schizophrenia group is that the CHR “converters” included some people with “trivial” transitions. Although there appears to be no underlying process in those with a "trivial" transition, there is, however, very poor neurocognitive functioning in some others who made transition, indicating development of schizophrenia, possibly with implications for poor outcome.

References:

Wood SJ, Brewer W, Koutsouradis P, Phillips LJ, Francey SM, Proffitt T, Yung AR, Jackson HJ, McGorry PD, Pantelis C, Cognitive decline following psychosis onset: Data from the PACE clinic. British Journal of Psychiatry 2007; 191 (suppl 51):52-57. Abstract

Hawkins KA, Keefe RS, Christensen BK, et al., Neuropsychological course in the prodrome and first episode of psychosis: Findings from the PRIME North America Double Blind Treatment Study. Schizophrenia Research 2008; 105:1-9. Abstract

Yung AR, Nelson B, Thompson A, Wood SJ, The psychosis threshold in Ultra High Risk (“prodromal”) research: is it valid? Schizophr Res 2010; 120:1-6. Abstract

View all comments by Alison Yung

When the eight centers participating in the NAPLES group decided to bring their own data gained from independently conducted trials into a large communal pool, the plan, although followed with great interest by other prodrome research centers worldwide, was also regarded as problematic. There is, in fact, a methodological difference in achieving a large database by pooling data a posteriori versus performing a prospective longitudinal multicenter research project such as the European Prediction of Psychosis Study (EPOS) (Ruhrmann et al., 2010). Previous NAPLES publications have convinced the scientific community, however, that the methodological problem of the disparate data collection can still be largely solved with the aid of appropriate integration procedures. The similarity of the ascertainment and the diagnostic methods made it possible to form a standardized protocol for mapping data into a new scheme representing the common components across sites. Through this careful processing, a solid database for the important development of the multivariate clinical NAPLES prediction model, which is already widely accepted by the scientific community, was established. Seidman et al. presents this methodological basis that confers its evidential power on the analysis of the jointly collected neuropsychological components of all participating centers.

After the successful integration of the clinical data and the effective development of the multivariate NAPLES clinical algorithm, it was only logical to apply the same procedure to the neuropsychological data. Doing so, the authors pursued the hypotheses that persons who subsequently convert to psychosis are more impaired at baseline than non-converters, that this effect would be amplified in persons with a family history (FH) of psychosis, and that poorer neuropsychological performance would predict a more rapid conversion to psychosis. The first two hypotheses were clearly confirmed, and the third was confirmed by all but one of the eight neurocognitive variables, namely, verbal memory.

The authors correctly state that their study is largely consistent with the results of other research groups, including those comparing converters with non-converters. Their major conclusion that processing speed and verbal learning/memory are most sensitive in discriminating clinical high-risk individuals from normal controls has been stated by several authors before, particularly by authors of other longitudinal studies. Lencz et al. (2006) found declarative verbal memory to be predictive for conversion to psychosis. Pukrop et al. (2007) identified verbal memory and processing speed (primarily the Digit Symbol Coding). Similarly, Riecher-Rössler et al., (2009) again reported a processing speed measure (a Go/No go test) to be the best neuropsychological predictor for conversion. Thus, exactly the same set of predictors postulated by the NAPLES group has already been identified by previous longitudinal investigations. Moreover, a recent review on the majority of all clinical HR studies published so far came to the same conclusion—processing speed and verbal memory are the most valid predictors (Pukrop and Klosterkötter, 2010).

The authors also rightly emphasize that the particular strengths of the study were the large sample size, achieved by the data pooling, as well as the relatively long follow-up of up of two and a half years. However, only 167 of the original 304 clinical high-risk subjects (CHR) were included in the central multivariate analyses with follow-up data, and also the second-biggest sample with published follow-up data still comprised 83 CHR-subjects (Pukrop et al., 2007). In addition, other longitudinal studies also report on follow-ups from six months up to five years (Lencz et al., 2006), with a median interval of three years (Pukrop et al. 2007), or follow-ups from 10 months to seven years (Riecher-Rössler et al., 2009). Also, some of the conclusions, such as those that state that neuropsychological deficits are intermediate between control and first-episode psychosis samples, should perhaps be considered more cautiously. There are a substantial number of studies reporting no significant differences between CHR subjects and patients with (first-episode) schizophrenia across a wide range of neuropsychological functions (e.g., Carr et al., 2000; Eastvold et al., 2007; Francey et al., 2005; Keefe et al., 2006; Koethe et al., 2009; Pukrop et al., 2006; Simon et al., 2007). Certainly, there is even more evidence providing significant differences between CHR subjects and patients with schizophrenia; however, the evidence is not as consistent as the authors suggested. The same also applies to the authors' statement that several specific deficits have been observed most reliably in spatial and verbal working memory, verbal declarative memory, and attention. This may be right for the functions of verbal working and declarative memory and attention, but can hardly be stated for spatial working memory. In the case of spatial working memory, there is less positive evidence (Wood et al., 2003; Smith et al., 2006; Chung et al., 2008) than negative evidence of missing abnormalities in CHR subjects (Hawkins et al., 2004; Keefe et al., 2006; Pukrop et al., 2006; Parnas et al., 2001). Moreover, what is not mentioned in the authors’ summary of former cross-sectional results is the fact that measures of processing speed such as the Digit Symbol Coding, a Go/Nogo test, or the Trail Making Test B, have already produced the most convincing evidence within neuropsychological batteries.

Overall, one can only congratulate the North American colleagues for having achieved such a broad and convincing confirmation of the previous state of knowledge in this field of research, along with the additional evaluation of their integrated database. There cannot be any doubt about the impairment of the CHR individuals due to significant neuropsychological difficulties, and the significance of those impairments for the development of psychotic disorders. However, the most interesting and important contribution the authors have made with this publication is not so much to be seen in the positive result of the successful replication and confirmation; it is much more in the only negative result of this study—that neuropsychological functioning did not contribute uniquely to the prediction of psychosis beyond clinical criteria. When the eight neurocognitive variables or the composite scores were added separately to the multivariate NAPLES prediction algorithms, none of the prediction statistics for the neurocognitive variables was significant.

With regard to the ongoing debate on the possible inclusion of a risk for psychosis syndrome, or attenuated psychotic symptoms syndrome, or a psychotypal disorder (Ruhrmann et al., 2010) in the DSM-V and ICD-11 revisions, the lack of significance in the predictive value of the neurocognitive variables might disappoint all working groups, many of whom intend to increase the predictive power of the clinical criteria via neurobiological criteria. Also for the upcoming new version of the schizophrenia diagnosis, it was indeed repeatedly postulated to combine the novel neurocognitive and other neurobiological findings, e.g., the neuroimaging, endophenotype, or genetic markers, with the clinical criteria. In mild cognitive impairment (MCI), this has been achieved. For the risk syndrome, additional specific neuroimaging and biochemical markers increase the predictive power for Alzheimer's dementia. But, if the neuropsychological impairment in CHR individuals does not uniquely contribute to the prediction of conversion, and none of the previously proposed biomarkers have sufficient diagnostic relevance to schizophrenia (Gaebel and Zielasek, 2008), then what does it mean for future research?

This question is, of course, too difficult to be adequately discussed in such a comment. Nevertheless, it ought to be disputed by all the working groups concerned with schizophrenia and other psychotic disorders. It could be that we still do not sufficiently characterize the underlying pathophysiology via our current neurobiological paradigms. It could also be the case that different kinds of pathophysiology are driving the clinical risk and diagnostic criteria, and we have not yet discerned them. Larry J. Seidman and the NAPLES group have not made such far-reaching considerations. Similar to the work in our group, future studies ought to selectively take the neuropsychological functional impairments into account that were associated with conversion to psychosis, and then examine large samples over a long period of time during which subsequent conversions could be recorded (Klosterkötter et al., 2001). These future proceedings will make it possible to clarify the potential for neuropsychological tests to function as early indictors of clinical conversion, as well as tools within CHR research.

References:

Carr V, Halpin S, Lau N, O'Brien S, Beckmann J, Lewin T.A risk factor screening and assessment protocol for schizophrenia and related psychosis. Aust N Z J Psychiatry. 2000 Nov;34 Suppl:S170-80. Abstract

Chung YS, Kang DH, Shin NY, Yoo SY, Kwon JS. Deficit of theory of mind in individuals at ultra-high-risk for schizophrenia. Schizophr Res 2008;99:111-118. Abstract

Eastvold AD, Heaton RK, Cadenhead KS. Neurocognitive deficits in the (putative) prodrome and first episode of psychosis. Schizophr Res. 2007;93(1-3):266-277. Abstract

Francey SM, Jackson HJ, Phillips LJ, Wood SJ, Yung AR, McGorry PD. Sustained attention in young people at high risk of psychosis does not predict transition to psychosis. Schizophr Res. 2005;79(1):127-136. Abstract

Gaebel W, Zielasek J. The DSM-V initiative “deconstructing psychosis” in the context of Kraepelin’s concept on nosology. Eur Arch Psychiatry Clin Neurosci. 2008258 (Suppl2):41-47. Abstract

Hawkins KA, Addington J, KeefeRS, Christensen B, PerkinsDO, ZipurksyR,Woods SW,MillerTJ,MarquezE,BreierA,McGlashanTH. Neuropsychological status of subjects at high risk for a first episode of psychosis. Schizophr Res 2004;67:115-122. Abstract

Keefe RSE, Perkins DO, Gu H, Zipursky RB, Christensen BK, Lieberman JA. A longitudinal study of neurocognitive function in individuals at-risk for psychosis. Schizophr Res. 2006;88(1-3):26-35. Abstract

Klosterkötter J, Hellmich M, Steinmeyer EM, Schultze-Lutter F: Diagnosing schizophrenia in the initial prodromal phase. Arch.Gen.Psychiatry 2001;58:158-164. Abstract

Koethe D, Kranaster L, Hoyer C, Gross S, Neatby MA, Schultze-Lutter F, Ruhrmann S, Klosterkötter J, Hellmich M, Leweke FM. Binocular depth inversion as a paradigm of reduced visual information processing in prodromal state, antipsychotic-naïve and treated schizophrenia. Eur Arch Psychiatry Clin Neurosci 2009 Jan 22. Abstract

Lencz T, Smith CW, McLaughlin D, Auther A, Nakayama E, Hovey L, Cornblatt BA. Generalized and specific neurocognitive deficits in prodromal schizophrenia. Biol Psychiatry, 2006;59(9):863-871. Abstract

Parnas, J., Vianin, P., Saebye, D., Jansson, L., Volmer-Larsen, A., Bovet, P., 2001. Visual binding abilities in the initial and advanced stages of schizophrenia. Acta Psychiatr. Scand. 103, 171-180. Abstract

Pukrop R, Schultze-Lutter F, Ruhrmann S, Brockhaus-Dumke A, Tendolkar I, Bechdolf A, Matuschek E, Klosterkötter J. Neurocognitive functioning in subjects at risk for a first episode of psychosis compared with first- and multiple-episode schizophrenia. J Clin Exp Neuropsychol 2006;28 (8):1388-407. Abstract

Pukrop R, Ruhrmann S, Schultze-Lutter F, Bechdolf A, Brockhaus-Dumke A, Klosterkötter J. Neurocognitive indicators for a conversion to psychosis: comparison of patients in a potentially initial prodromal state who did or did not convert to a psychosis. Schizophr Res 2007;92 (1-3):116-25. Abstract

Pukrop R, Klosterkötter J. Neurocognitive indicators of clinical high-risk states for psychosis: a critical review of the evidence. Neurotox Res. Abstract

Riecher-Rössler A, Pflueger MO, Aston J, Borgwardt SJ, Brewer WJ, Gschwandtner U, Stieglitz RD. Efficacy of using cognitive status in predicting psychosis: a 7-year follow-up. Biol Psychiatry. 2009 Dec 1;66(11):1023-30. Abstract

Ruhrmann S, Schultze-Lutter F, Salokangas RK, Heinimaa M, Linszen D, Dingemans PM, Birchwood M, Patterson P, Juckel G, Heinz A, Morrison A, Lewis S, von Reventlow HG, Klosterkötter J. Prediction of psychosis in adolescents and young adults at high risk: results from the prospective European Prediction of Psychosis Study. Arch Gen Psychiatry 2010 Mar;67(3):241-51. Abstract

Ruhrmann S, Schultze-Lutter F, Klosterkötter J. Probably at-risk, but certainly ill--advocating the introduction of a psychosis spectrum disorder in DSM-V. Schizophr Res. 2010 Jul;120(1-3):23-37. Abstract

Simon AE, Cattapan-Ludewig K, Zmilacher S, Arbach D, Gruber K, Dvorsky DN, Roth B, Isler E, Zimmer A, Umbricht D. Cognitive functioning in the schizophrenia prodrome. Schizophr Bull. 2007 May;33(3):761-71. Abstract

Smith CW, Park S, Cornblatt B. Spatial working memory deficits in adolescents at clinical high risk for schizophrenia. Schizophr Res. 2006;81(2-3):211-215. Abstract

Wood SJ, Pantelis C, Proffitt T, Phillips LJ, Stuart GW, Buchanan J, Mahony K, Brewer WJ, Smith D, McGorry PD. Spatial working memory ability is a marker of risk-for-psychosis. Psychol Med. 2003;33(7):1239-1247. Abstract

View all comments by Joachim Klosterkötter
View all comments by Ralf Pukrop