- Email: [email protected]
Pilot testing of the computerized cognitive test Microcog™ in chemotherapy-treated older cancer patients
Critical Reviews in Oncology/Hematology 54 (2005) 137–143
Pilot testing of the computerized cognitive test MicrocogTM in chemotherapy-treated older cancer patients Martine Extermann∗ , Hongbin Chen, Margaret Booth-Jones, Julie Meyer, Lodovico Balducci, Paul Jacobsen Senior Adult Oncology, H. Lee Moffitt Cancer Center, University of South Florida, 12902 Magnolia Drive, Tampa, FL 33612, USA Accepted 7 November 2004
Contents Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1. Statistical and ethical considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1. Feasibility issues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Reviewers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Biographies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1. 2. 3. 4.
137 138 138 138 141 142 142 143 143
Abstract Background: Chemotherapy has a potential for inducing cognitive side effects. However, no study has focused on elderly cancer patients, a group that might be at risk for this complication. Computerized cognitive tests are available and could simplify cooperative group studies on the matter, but have not been applied to older cancer patients. Methods: We tested the performance of MicrocogTM (short form) in a sample of 10 consecutive cancer patients, aged 70 and older, having received chemotherapy. Patients were also asked by questionnaire to express their comments on the test. Results: Six patients had never used a computer. All reported at least minor visual impairment. All did complete the test without pause. Nine out of 10 thought that most patients like them would have no problems completing the test. As a group, our patient sample generally performed within normal limits for age and education. There were a wide range of scores for the majority of the subscales, with the greatest variability of scores in Spatial Processing and Information Processing Accuracy and the least variability in reaction time. The results were robust when assessed by level of computer literacy, minor auditory and visual problems, and fluent English as a second language. Conclusions: A computer test such as MicrocogTM appears well feasible in older cancer patients. It appears robust to comorbidity. This bodes well for a potential use of such tests in trials conducted in this patient population. © 2004 Elsevier Ireland Ltd. All rights reserved. Keywords: Aged; Aged > 80; Cognitive assessment; Microcog; Chemotherapy
1. Introduction
∗
Corresponding author. Tel.: +1 813 979 3822; fax: +1 813 972 8359. E-mail address: [email protected] (M. Extermann).
1040-8428/$ – see front matter © 2004 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.critrevonc.2004.11.009
There is increasing evidence that standard dose chemotherapy can induce mild albeit clinically significant cognitive impairment in 17–75% of adult patients, depend-
138
M. Extermann et al. / Critical Reviews in Oncology/Hematology 54 (2005) 137–143
ing on the regimen and site of tumor [1–6]. Specific information about older patients is lacking. This is especially of concern for people with subclinical or mild preexisting dementia, which could deteriorate to a clinically significant level as a result of treatment. This could also impair further informed consent to the treatment process in approaches using sequential treatments. It is unknown whether older patients experience a decline proportionate to that seen in young patients, or whether underlying brain aging leads to worse impact of chemotherapy. It is therefore crucial to investigate this information in groups of patients large enough for detailed analysis. In order to conduct these trials in large cooperative group studies, it is important to have an assessment tool presenting the following qualities: high sensitivity and variability at a mild level of cognitive impairment, ability to be used by research associates in many hospitals with minimal training, good acceptance by the patients, good interrater and test–retest reliability, reasonable duration and cost of the test. Several tools are available for cognitive assessment in this population, such as the Mattis Dementia Rating Scale [7] and the Repeatable Battery for the Assessment of Neuropsychological Status [8]. However, most are interviewer administered, thus requiring more personnel resources and add to the variation in standardization of administration. Recently, computer administered neuropsychological screening tools have been developed. Their theoretical advantages are a better standardization of administration in a multicentric setting, and the ability for the patients to proceed through the testing with minimal help from a clinical research associate. One of them is MicrocogTM . MicrocogTM : Assessment of Cognitive Functioning, Version 2.4, assesses nine areas of cognition, specifically attention/mental control, information processing, reaction time, accuracy, memory, information processing speed, reasoning/calculation, spatial processing, and cognitive proficiency. Normative data is available for ages 18–89, with specific norms for age and education level [9]. The standard form takes 50–60 min for administration, however, the short form requires approximately 30 min. Scoring and interpretation are performed by the accompanying software [10]. We tested the feasibility and acceptance of MicrocogTM by older cancer patients in an outpatient clinical setting: Moffitt’s Senior Adult Oncology Program (SAOP). The SAOP is focused on a comprehensive approach for patients aged 70 years and older.
2. Methods A successive cohort of patients from the SAOP, aged 70 and older, all receiving or having received chemotherapy were asked to complete the MicrocogTM assessment at the occasion of a clinic or a treatment visit. Patients with major vision or hearing problems preventing the use of a computer or with an established clinical diagnosis of dementia (DSM IV) were excluded. Age, gender, type and stage of tumor,
past and present treatment, education level, hearing impairment (per history) with or without correction, visual impairments (per history) with or without correction were recorded. MicrocogTM was loaded onto a laptop computer installed in a quiet room near the clinic. The test was explained and monitored by an investigator (HC), according to the instructions manual. The time to complete the test was recorded. At the end, each patient was asked to fill a questionnaire about the ease of use of the program. The MicrocogTM software provides a standardized rating in each of the nine domains tested. This means that the mean population score is attributed the value of 100 points, and the standard deviation a value of 15 points. The score is standardized either by comparison to young adults, or by comparison to an age/education-adjusted sample. Higher scores reflect better cognitive function. The short form comprises 12 subsets: timers, address presentation, story 1 immediate recall, clocks, story 2 immediate recall, numbers forward, analogies, wordlist 1 and 2, address recall, math, story 1 delayed, story 2 delayed. The following end-points were assessed: total number of participants who were able to complete MicrocogTM . If they did not complete it, the reason for non-completion was recorded. Completion of MicrocogTM without missing items was another end-point, as well as the time to complete the assessment. Patient answers to the questionnaire about ease of use were evaluated. Finally, we wanted to assess the range of scores generated in this population, especially potential floor and ceiling effects. This would allow us to assess whether MicrocogTM would have a potential to capture variability in functioning in patients expected for the most part to have subtle rather than gross cognitive impairments [1]. 2.1. Statistical and ethical considerations The evaluation was designed as mainly descriptive. Pearson correlation analyses were conducted on age and education versus MicrocogTM scores. A clinical judgement about the feasibility and acceptability of the software was made. This protocol was approved by the IRB of the University of South Florida. Patients signed an informed consent prior to the study.
3. Results Ten patients were accrued, three men and seven women. The age range was 71–85 years, with a mean of 78. The mean level of education was 14.1 years, with a range of 5–20. All patients were right handed. Three patients had English as a second language. All were fluent in English. The cancer diagnoses were colon cancer (3), breast cancer (3), esophageal cancer (1) (with a past history of lymphoma and prostate cancer), ovarian cancer (1), non-Hodgkin’s lymphoma (1), and chronic lymphocytic leukemia (1). Six were undergoing chemotherapy and four
M. Extermann et al. / Critical Reviews in Oncology/Hematology 54 (2005) 137–143
had finished it. The treatments received included concomitant radiation-5-FU, 5-FU/leucovorin, oxaliplatin, irinotecan, CAF, AC-taxol, vinorelbine, capecitabine, CHOP, concomitant carboplatin–paclitaxel-radiation, carboplatin–paclitaxel, carboplatin, Doxil, chlorambucil, and fludarabine. Two patients were undergoing their first line of chemotherapy, four patients had received one line of treatment, one patient had received two lines of treatment, two patients were receiving their third-line chemotherapy, and one patient had received four lines of chemotherapy. Six patients reported never having used a computer, one was using a computer daily, two weekly, and one less than once a week. All patients reported visual problems: nine minor, one major, all with good correction. Four patients reported hearing problems: three minor, one major, one with poor correction. All patients were able to complete the test, and none used the pause option. The time for completion was 30–45 min, median 35 min. The results in each domain are shown in Table 1. The software uses the number of years of education as an adjustment variable. No independent correlation was found, and this supports the validity of the internal adjustment of the software. Age correlated significantly (p < 0.05) with reasoning/calculation for both reference norms and age/education corrected norms (Table 1). Age also correlated significantly (p < 0.05) with information processing speed for age/education corrected norms. This might point to a differ-
139
Table 1 Results from MicrocogTM test Subtest
Mean
S.D.
Range
Percentage
MicrocogTM :
index scaled scores—age and education corrected norms Attention/mental control 94.1 18.7 66–121 20 Reasoning/calculation 98.8 13.6 74–115 0 Memory 103.7 18.1 81–138 0 Spatial processing 108.1 22.7 54–126 10 Reaction time 99.9 10.9 84–120 0 Information processing speed 88.3 14.9 64–104 10 Information processing accuracy 106.2 20.7 69–148 10 General cognitive functioning 97.1 19.4 63–130 10 General cognitive proficiency 94.5 13.4 76–121 0
This represents the performance of the whole group of patients compared to same age reference subjects. The last column represents the percentage of patients that fall below a score of 70, or two standard deviations below norm.
ential toxicity of chemotherapy with age. The impact on the results of hearing and visual impairment, having English as a second language, and computer literacy is shown in Figs. 1–4. Neither of these factors did obviously alter the results of the test in this small sample. Specific subscale scores are presented in Table 1. Results indicate that as a group the sample generally performed within normal limits for age and education. Closer analysis reveals a wide range of scores for the majority of the subscales, with the greatest variability of age and educationadjusted scores in spatial processing and information pro-
Fig. 1. Individual patient scores with the influence of impaired hearing (dotted lines). Scores standardized for age.
140
M. Extermann et al. / Critical Reviews in Oncology/Hematology 54 (2005) 137–143
Fig. 2. Individual patient scores with the influence of major visual impairment, corrected (dotted line).
Fig. 3. Individual patient scores with the influence of computer literacy (dotted lines).
M. Extermann et al. / Critical Reviews in Oncology/Hematology 54 (2005) 137–143
141
Fig. 4. Individual patient scores with the influence of English as a second language (dotted lines).
cessing accuracy and the least variability in reaction time. Using a score of two standard deviations below the mean (standard score of 70 or less) to define impairment, two participants performed in the range of impairment on attention/mental control, and one participant performed in the range of impairment on spatial processing, information processing speed, information processing accuracy and general cognitive functioning. None of the participants performed in the range of impairment on the following subtests: reasoning/calculation, memory, reaction time, and general cognitive proficiency. One participant accounted for five of the six impaired subtest scores. Patients were very comfortable with the test, and 9 out of 10 thought that patients like them would have no problems completing the test. They rated the ease of understanding the instructions, reading the written and hearing the verbal questions, and using the laptop as good to very good (3.1–3.4 on a scale from 1 to 5, 5 being the best). They rated the duration of the test as fair to good (2.7). Subjective written comments showed good acceptance, with several comments suggesting that the test offered “a fun challenge”.
4. Discussion Reported studies of the cognitive impact of chemotherapy, using various testing strategies, describe a broad spectrum of
alterations. In one study attention, mental flexibility, speed of information processing, visual memory, and motor function were affected [2]. In another study, language, attention and concentration, self-regulation and planning tended to be worse in patients [4]. In a study on the cognitive impact of androgen suppression, the authors found a decline in measures of spatial rotation and an improvement in verbal memory [3]. Overall, the impact appears fairly diffuse, but memory and executive functions could be preferentially affected [5]. These studies were for the vast majority using batteries of tests administered directly by neuropsychologists [6]. How well does MicrocogTM address these domains, and how sensitive is it to changes in the range of chemotherapyinduced cognitive changes? Several studies did compare MicrocogTM to full neurological testing in various settings. It appears to correlate well with a traditional battery of tests, with a trend to rate patients more severely [11]. Extensive validity information is available from studies of clinical groups including epilepsy, depression, substance abuse, gulf war veterans and dementia patients [10–14]. MicrocogTM was used successfully to distinguish older adults with and without mild cognitive impairment. Using MicrocogTM (then called the assessment of cognitive skills (ACS)), Green and colleagues [13] compared cognitive functioning in a sample of older adults (mean age 71 years) exhibiting mild cognitive impairment (mostly probable Alzheimer’s by National Institute of Neurological and Communicative Disorders criteria) and
142
M. Extermann et al. / Critical Reviews in Oncology/Hematology 54 (2005) 137–143
an age- and education-matched comparison group of older adults with no evidence of cognitive impairment after a thorough neurological screening. Findings indicated the mean ACS total score for controls was significantly better than the mean for the mildly impaired. Moreover, use of a cutoff score to identify cognitive impairment resulted high rates of sensitivity (0.83) and specificity (0.96) for distinguishing the two groups. Since most cognitive impairment from chemotherapy is mild to moderate, it appears from these data that MicrocogTM has a promising potential in that range of scores. MicrocogTM also appears relatively unaffected by depression, an important point when considering its use in cancer patients [10]. We used the short form of the MicrocogTM test, in order to lessen the burden in cancer patients. This form correlates well with the standard form, with the exception of the spatial score, where the standard form appears more demanding than the short form [14]. Given the domains touched in the studies mentioned above, despite its lower sensitivity in the spatial score, the short form of MicrocogTM appears a reasonable choice in cancer patients. Our sample of chemotherapytreated patients, despite its small size, did not highlight any trend toward a floor or a ceiling effect. The wide range of scores obtained is promising for the test’s availability to detect variations. Two patients scored below the lower limit of normal (−2 S.D. or 70 points). One had subjective complaints corresponding to the test findings. He had finished a chemoradiation therapy 1 month earlier. He qualified his problems as minor and improved later clinically. Another patient scored low on several subscales (see figures). This patient on further interview disclosed a diagnosis of dyslexia limited to calculations. Its role in the scoring is unclear, and patient declined further cognitive assessment. We suggest that information about dyslexia should be collected prospectively in future studies. MicrocogTM is not the only computerized test available, only the oldest one. A more recent test called cognitive stability index (CSI) has a more convenient computer interface. Due to its novelty, only one study using was published at the time of this study [15]. More literature is being published. For example, it has been successfully able to distinguish between independent and mildly dependent older adults [16]. A third test is the cognitive drug research (CDR) test. A larger body of literature is available (e.g. [17]). However, its cost is much higher than that of the two previous tests. Computerized memory testing is attractive in many settings and the list of available programs keeps expanding. 4.1. Feasibility issues MicrocogTM proved a very usable test in our cohort of older chemotherapy-treated patients. The program appears to adjust well for education. However, confounding factors might be present: for example, the patient with the least formal education was also the one speaking three languages. Therefore a more direct measurement of literacy, such as the
North American Reading Test, might be helpful in interpreting the results. Overall, our group of patients was well functioning. Testing in a larger sample will be needed to assess whether the discriminative power of MicrocogTM is the same as in older non-cancer patients [13]. A major concern that prompted this feasibility study was to ensure that computer illiterate patients, or patients with visual or auditory impairments such as those encountered in our usual population of elderly cancer patients could complete the test easily. Most of our patients were computer illiterate, yet performed very well and enjoyed the test. Corrected visual impairments did not appear to interfere with the test. Similarly, the three patients with English as a second language scored similarly to the others. This test has some downsides. These are a somewhat difficult conversion into the SAS statistical software, which we corrected with a few lines of software, and the cost (from $ 6 to 10 per test, plus $ 250 for the program). However, one should note that paper tests cost about $ 5 per test. We believe that the gains in reproducibility, personnel needs, and patient acceptability make the investment worthwhile. In conclusion, MicrocogTM is a very feasible test in elderly cancer patients having received chemotherapy, with promising properties for studies in this patient population. This pilot also demonstrated in a more general way the feasibility of computerized cognitive testing in this population. Other computer tests of more modern design should be assayed as well, and their performance compared in this patient population. It is at this point too early to propose any direct clinical use of these tests in older cancer patients. Another challenge is the increasing number of patients with established dementia who may present a cancer. Little is known about treatment patterns and prognosis of these patients, even more about the cognitive impact of cancer treatment. MicrocogTM has been validated in dementia patients [13]. Its usefulness in cancer patients with established dementia remains to be evaluated. Reviewers Carol M. Moinpour, Ph.D., Behavioral Scientist/Southwest Oncology Group Statistical Center, Associate Member/Division of Public Health Sciences, Fred Hutchinson Cancer Research Center, 1100 Fairview Av. North/MP-557, Box 19024, Seattle, WA 98109-1024, USA. Pand´elis Giannakopoulos, M.D., FMH (Psy), Professor, Associate Vice-Dean, Service of Geriatric Psychiatry, Department of Psychiatry-HUG-Belle-Idee, Pavilion Les Champs, 2 Ch. du Petit-Bel-Air, CH-1225 Chˆene-Bourg, Geneva; Psychogeriatric Service, University of Lausanne, CH-1008 Prilly, Switzerland. Dr. Gilbert Zulian, Centre de Soins Continus (CESCO), Ch. de la Savonni`ere 11, CH-1245 Collonge-Bellerive, Switzerland.
M. Extermann et al. / Critical Reviews in Oncology/Hematology 54 (2005) 137–143
References [1] Ahles TA, Saykin A. Cognitive effects of standard-dose chemotherapy in patients with cancer. Cancer Invest 2001;19:812–20. [2] Schagen SB, van Dam FS, Muller MJ, Boogerd W, Lindeboom J, Bruning PF. Cognitive deficits after postoperative adjuvant chemotherapy for breast carcinoma. Cancer 1999;85:640– 50. [3] Cherrier MM, Rose AL, Higano C. The effects of combined androgen blockade on cognitive function during the first cycle of intermittent androgen suppression in patients with prostate cancer. J Urol 2003;170:1808–11. [4] Tchen N, Juffs HG, Downie FP, et al. Cognitive function, fatigue, and menopausal symptoms in women receiving adjuvant chemotherapy for breast cancer. J Clin Oncol 2003;21:4175–83. [5] Saykin AJ, Ahles TA, McDonald BC. Mechanisms of chemotherapyinduced cognitive disorders: neuropsychological, pathophysiological, and neuroimaging perspectives. Semin Clin Neuropsychiatry 2003;8:201–16. [6] Tannock IF, Ahles TA, Ganz PA, Van Dam FS. Cognitive impairment associated with chemotherapy for cancer: report of a workshop. J Clin Oncol 2004;22:2233–9. [7] Mattis S. Dementia rating scale. Odessa, FL: Psychological Assessment Resources; 1989. [8] Randolph C. RBANS manual—repeatable battery for the assessment of neuropsychological status. San Antonio, TX: The Psychological Corporation, Harcourt; 1998. [9] Powell DH, Kaplan EF, Whitla D, Weintraub S, Catlin R, Funkenstein HH. MicrocogTM manual. San Antonio, TX: The Psychological Corporation, Harcourt Brace & Co.; 1993. [10] Elwood RW. MicroCog: assessment of cognitive functioning. Neuropsychol Rev 2001;11:89–100. [11] Holliday SL, Costello RM. Screening for neuropsychological deficits in Gulf War veterans: MicroCog vs standard neuropsychological tests. Arch Clin Neuropsychol 1997;12:338. [12] Moore JL, McAuley JW, Long L, Bornstein R. An evaluation of the effects of methylphenidate on outcomes in adult epilepsy patients. Epilepsy Behav 2002;3:92–5. [13] Green RC, Green J, Harrison JM, Kutner MH. Screening for cognitive impairment in older individuals—validation study of a computerbased test. Arch Neurol 1994;51:779–86.
143
[14] Lopez SJ, Edwards LM, Floyd RK, Magyar-Moe J, Rehfeldt JD, Ryder JA. Note on comparability of Microcog test forms. Percept Motor Skills 2001;93:825–8. [15] Erlanger DM, Kaushik T, Broshek D, Freeman J, Feldman D, Festa J. Development and validation of a web-based screening tool for monitoring cognitive status. J Head Trauma Rehab 2002;17: 458–76. [16] Petrella JK, Miller LS, Cress ME. Leg extensor power, cognition, and functional performance in independent and marginally dependent older adults. Age and Ageing 2004;33:342–8. [17] Wesnes KA, McKeith IG, Ferrara R, et al. Effects of rivastigmine on cognitive function in dementia with lewy bodies: a randomised placebo-controlled international study using the cognitive drug research computerised assessment system. Dement Geriatr Cogn Disorder 2002;13:183–92.
Biographies Martine Extermann is Associate Professor in the Senior Adult Oncology Program (SAOP) at H. Lee Moffitt Cancer Center, University of South Florida (USF). Hongbin Chen was Assistant Director of the Florida Statistical Center on Aging at USF during the study. Margaret Booth-Jones is a neuropsychologist in the Psychosocial and Palliative Care Program at H. Lee Moffitt Cancer Center. Julie Meyer is the nurse practitioner of the SAOP. Lodovico Balducci is Professor and Program Leader of the SAOP. Paul Jacobsen is a psychologist and Program Leader of Program.
Pilot testing of the computerized cognitive test MicrocogTM in chemotherapy-treated older cancer patients Martine Extermann∗ , Hongbin Chen, Margaret Booth-Jones, Julie Meyer, Lodovico Balducci, Paul Jacobsen Senior Adult Oncology, H. Lee Moffitt Cancer Center, University of South Florida, 12902 Magnolia Drive, Tampa, FL 33612, USA Accepted 7 November 2004
Contents Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1. Statistical and ethical considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1. Feasibility issues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Reviewers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Biographies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1. 2. 3. 4.
137 138 138 138 141 142 142 143 143
Abstract Background: Chemotherapy has a potential for inducing cognitive side effects. However, no study has focused on elderly cancer patients, a group that might be at risk for this complication. Computerized cognitive tests are available and could simplify cooperative group studies on the matter, but have not been applied to older cancer patients. Methods: We tested the performance of MicrocogTM (short form) in a sample of 10 consecutive cancer patients, aged 70 and older, having received chemotherapy. Patients were also asked by questionnaire to express their comments on the test. Results: Six patients had never used a computer. All reported at least minor visual impairment. All did complete the test without pause. Nine out of 10 thought that most patients like them would have no problems completing the test. As a group, our patient sample generally performed within normal limits for age and education. There were a wide range of scores for the majority of the subscales, with the greatest variability of scores in Spatial Processing and Information Processing Accuracy and the least variability in reaction time. The results were robust when assessed by level of computer literacy, minor auditory and visual problems, and fluent English as a second language. Conclusions: A computer test such as MicrocogTM appears well feasible in older cancer patients. It appears robust to comorbidity. This bodes well for a potential use of such tests in trials conducted in this patient population. © 2004 Elsevier Ireland Ltd. All rights reserved. Keywords: Aged; Aged > 80; Cognitive assessment; Microcog; Chemotherapy
1. Introduction
∗
Corresponding author. Tel.: +1 813 979 3822; fax: +1 813 972 8359. E-mail address: [email protected] (M. Extermann).
1040-8428/$ – see front matter © 2004 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.critrevonc.2004.11.009
There is increasing evidence that standard dose chemotherapy can induce mild albeit clinically significant cognitive impairment in 17–75% of adult patients, depend-
138
M. Extermann et al. / Critical Reviews in Oncology/Hematology 54 (2005) 137–143
ing on the regimen and site of tumor [1–6]. Specific information about older patients is lacking. This is especially of concern for people with subclinical or mild preexisting dementia, which could deteriorate to a clinically significant level as a result of treatment. This could also impair further informed consent to the treatment process in approaches using sequential treatments. It is unknown whether older patients experience a decline proportionate to that seen in young patients, or whether underlying brain aging leads to worse impact of chemotherapy. It is therefore crucial to investigate this information in groups of patients large enough for detailed analysis. In order to conduct these trials in large cooperative group studies, it is important to have an assessment tool presenting the following qualities: high sensitivity and variability at a mild level of cognitive impairment, ability to be used by research associates in many hospitals with minimal training, good acceptance by the patients, good interrater and test–retest reliability, reasonable duration and cost of the test. Several tools are available for cognitive assessment in this population, such as the Mattis Dementia Rating Scale [7] and the Repeatable Battery for the Assessment of Neuropsychological Status [8]. However, most are interviewer administered, thus requiring more personnel resources and add to the variation in standardization of administration. Recently, computer administered neuropsychological screening tools have been developed. Their theoretical advantages are a better standardization of administration in a multicentric setting, and the ability for the patients to proceed through the testing with minimal help from a clinical research associate. One of them is MicrocogTM . MicrocogTM : Assessment of Cognitive Functioning, Version 2.4, assesses nine areas of cognition, specifically attention/mental control, information processing, reaction time, accuracy, memory, information processing speed, reasoning/calculation, spatial processing, and cognitive proficiency. Normative data is available for ages 18–89, with specific norms for age and education level [9]. The standard form takes 50–60 min for administration, however, the short form requires approximately 30 min. Scoring and interpretation are performed by the accompanying software [10]. We tested the feasibility and acceptance of MicrocogTM by older cancer patients in an outpatient clinical setting: Moffitt’s Senior Adult Oncology Program (SAOP). The SAOP is focused on a comprehensive approach for patients aged 70 years and older.
2. Methods A successive cohort of patients from the SAOP, aged 70 and older, all receiving or having received chemotherapy were asked to complete the MicrocogTM assessment at the occasion of a clinic or a treatment visit. Patients with major vision or hearing problems preventing the use of a computer or with an established clinical diagnosis of dementia (DSM IV) were excluded. Age, gender, type and stage of tumor,
past and present treatment, education level, hearing impairment (per history) with or without correction, visual impairments (per history) with or without correction were recorded. MicrocogTM was loaded onto a laptop computer installed in a quiet room near the clinic. The test was explained and monitored by an investigator (HC), according to the instructions manual. The time to complete the test was recorded. At the end, each patient was asked to fill a questionnaire about the ease of use of the program. The MicrocogTM software provides a standardized rating in each of the nine domains tested. This means that the mean population score is attributed the value of 100 points, and the standard deviation a value of 15 points. The score is standardized either by comparison to young adults, or by comparison to an age/education-adjusted sample. Higher scores reflect better cognitive function. The short form comprises 12 subsets: timers, address presentation, story 1 immediate recall, clocks, story 2 immediate recall, numbers forward, analogies, wordlist 1 and 2, address recall, math, story 1 delayed, story 2 delayed. The following end-points were assessed: total number of participants who were able to complete MicrocogTM . If they did not complete it, the reason for non-completion was recorded. Completion of MicrocogTM without missing items was another end-point, as well as the time to complete the assessment. Patient answers to the questionnaire about ease of use were evaluated. Finally, we wanted to assess the range of scores generated in this population, especially potential floor and ceiling effects. This would allow us to assess whether MicrocogTM would have a potential to capture variability in functioning in patients expected for the most part to have subtle rather than gross cognitive impairments [1]. 2.1. Statistical and ethical considerations The evaluation was designed as mainly descriptive. Pearson correlation analyses were conducted on age and education versus MicrocogTM scores. A clinical judgement about the feasibility and acceptability of the software was made. This protocol was approved by the IRB of the University of South Florida. Patients signed an informed consent prior to the study.
3. Results Ten patients were accrued, three men and seven women. The age range was 71–85 years, with a mean of 78. The mean level of education was 14.1 years, with a range of 5–20. All patients were right handed. Three patients had English as a second language. All were fluent in English. The cancer diagnoses were colon cancer (3), breast cancer (3), esophageal cancer (1) (with a past history of lymphoma and prostate cancer), ovarian cancer (1), non-Hodgkin’s lymphoma (1), and chronic lymphocytic leukemia (1). Six were undergoing chemotherapy and four
M. Extermann et al. / Critical Reviews in Oncology/Hematology 54 (2005) 137–143
had finished it. The treatments received included concomitant radiation-5-FU, 5-FU/leucovorin, oxaliplatin, irinotecan, CAF, AC-taxol, vinorelbine, capecitabine, CHOP, concomitant carboplatin–paclitaxel-radiation, carboplatin–paclitaxel, carboplatin, Doxil, chlorambucil, and fludarabine. Two patients were undergoing their first line of chemotherapy, four patients had received one line of treatment, one patient had received two lines of treatment, two patients were receiving their third-line chemotherapy, and one patient had received four lines of chemotherapy. Six patients reported never having used a computer, one was using a computer daily, two weekly, and one less than once a week. All patients reported visual problems: nine minor, one major, all with good correction. Four patients reported hearing problems: three minor, one major, one with poor correction. All patients were able to complete the test, and none used the pause option. The time for completion was 30–45 min, median 35 min. The results in each domain are shown in Table 1. The software uses the number of years of education as an adjustment variable. No independent correlation was found, and this supports the validity of the internal adjustment of the software. Age correlated significantly (p < 0.05) with reasoning/calculation for both reference norms and age/education corrected norms (Table 1). Age also correlated significantly (p < 0.05) with information processing speed for age/education corrected norms. This might point to a differ-
139
Table 1 Results from MicrocogTM test Subtest
Mean
S.D.
Range
Percentage
MicrocogTM :
index scaled scores—age and education corrected norms Attention/mental control 94.1 18.7 66–121 20 Reasoning/calculation 98.8 13.6 74–115 0 Memory 103.7 18.1 81–138 0 Spatial processing 108.1 22.7 54–126 10 Reaction time 99.9 10.9 84–120 0 Information processing speed 88.3 14.9 64–104 10 Information processing accuracy 106.2 20.7 69–148 10 General cognitive functioning 97.1 19.4 63–130 10 General cognitive proficiency 94.5 13.4 76–121 0
This represents the performance of the whole group of patients compared to same age reference subjects. The last column represents the percentage of patients that fall below a score of 70, or two standard deviations below norm.
ential toxicity of chemotherapy with age. The impact on the results of hearing and visual impairment, having English as a second language, and computer literacy is shown in Figs. 1–4. Neither of these factors did obviously alter the results of the test in this small sample. Specific subscale scores are presented in Table 1. Results indicate that as a group the sample generally performed within normal limits for age and education. Closer analysis reveals a wide range of scores for the majority of the subscales, with the greatest variability of age and educationadjusted scores in spatial processing and information pro-
Fig. 1. Individual patient scores with the influence of impaired hearing (dotted lines). Scores standardized for age.
140
M. Extermann et al. / Critical Reviews in Oncology/Hematology 54 (2005) 137–143
Fig. 2. Individual patient scores with the influence of major visual impairment, corrected (dotted line).
Fig. 3. Individual patient scores with the influence of computer literacy (dotted lines).
M. Extermann et al. / Critical Reviews in Oncology/Hematology 54 (2005) 137–143
141
Fig. 4. Individual patient scores with the influence of English as a second language (dotted lines).
cessing accuracy and the least variability in reaction time. Using a score of two standard deviations below the mean (standard score of 70 or less) to define impairment, two participants performed in the range of impairment on attention/mental control, and one participant performed in the range of impairment on spatial processing, information processing speed, information processing accuracy and general cognitive functioning. None of the participants performed in the range of impairment on the following subtests: reasoning/calculation, memory, reaction time, and general cognitive proficiency. One participant accounted for five of the six impaired subtest scores. Patients were very comfortable with the test, and 9 out of 10 thought that patients like them would have no problems completing the test. They rated the ease of understanding the instructions, reading the written and hearing the verbal questions, and using the laptop as good to very good (3.1–3.4 on a scale from 1 to 5, 5 being the best). They rated the duration of the test as fair to good (2.7). Subjective written comments showed good acceptance, with several comments suggesting that the test offered “a fun challenge”.
4. Discussion Reported studies of the cognitive impact of chemotherapy, using various testing strategies, describe a broad spectrum of
alterations. In one study attention, mental flexibility, speed of information processing, visual memory, and motor function were affected [2]. In another study, language, attention and concentration, self-regulation and planning tended to be worse in patients [4]. In a study on the cognitive impact of androgen suppression, the authors found a decline in measures of spatial rotation and an improvement in verbal memory [3]. Overall, the impact appears fairly diffuse, but memory and executive functions could be preferentially affected [5]. These studies were for the vast majority using batteries of tests administered directly by neuropsychologists [6]. How well does MicrocogTM address these domains, and how sensitive is it to changes in the range of chemotherapyinduced cognitive changes? Several studies did compare MicrocogTM to full neurological testing in various settings. It appears to correlate well with a traditional battery of tests, with a trend to rate patients more severely [11]. Extensive validity information is available from studies of clinical groups including epilepsy, depression, substance abuse, gulf war veterans and dementia patients [10–14]. MicrocogTM was used successfully to distinguish older adults with and without mild cognitive impairment. Using MicrocogTM (then called the assessment of cognitive skills (ACS)), Green and colleagues [13] compared cognitive functioning in a sample of older adults (mean age 71 years) exhibiting mild cognitive impairment (mostly probable Alzheimer’s by National Institute of Neurological and Communicative Disorders criteria) and
142
M. Extermann et al. / Critical Reviews in Oncology/Hematology 54 (2005) 137–143
an age- and education-matched comparison group of older adults with no evidence of cognitive impairment after a thorough neurological screening. Findings indicated the mean ACS total score for controls was significantly better than the mean for the mildly impaired. Moreover, use of a cutoff score to identify cognitive impairment resulted high rates of sensitivity (0.83) and specificity (0.96) for distinguishing the two groups. Since most cognitive impairment from chemotherapy is mild to moderate, it appears from these data that MicrocogTM has a promising potential in that range of scores. MicrocogTM also appears relatively unaffected by depression, an important point when considering its use in cancer patients [10]. We used the short form of the MicrocogTM test, in order to lessen the burden in cancer patients. This form correlates well with the standard form, with the exception of the spatial score, where the standard form appears more demanding than the short form [14]. Given the domains touched in the studies mentioned above, despite its lower sensitivity in the spatial score, the short form of MicrocogTM appears a reasonable choice in cancer patients. Our sample of chemotherapytreated patients, despite its small size, did not highlight any trend toward a floor or a ceiling effect. The wide range of scores obtained is promising for the test’s availability to detect variations. Two patients scored below the lower limit of normal (−2 S.D. or 70 points). One had subjective complaints corresponding to the test findings. He had finished a chemoradiation therapy 1 month earlier. He qualified his problems as minor and improved later clinically. Another patient scored low on several subscales (see figures). This patient on further interview disclosed a diagnosis of dyslexia limited to calculations. Its role in the scoring is unclear, and patient declined further cognitive assessment. We suggest that information about dyslexia should be collected prospectively in future studies. MicrocogTM is not the only computerized test available, only the oldest one. A more recent test called cognitive stability index (CSI) has a more convenient computer interface. Due to its novelty, only one study using was published at the time of this study [15]. More literature is being published. For example, it has been successfully able to distinguish between independent and mildly dependent older adults [16]. A third test is the cognitive drug research (CDR) test. A larger body of literature is available (e.g. [17]). However, its cost is much higher than that of the two previous tests. Computerized memory testing is attractive in many settings and the list of available programs keeps expanding. 4.1. Feasibility issues MicrocogTM proved a very usable test in our cohort of older chemotherapy-treated patients. The program appears to adjust well for education. However, confounding factors might be present: for example, the patient with the least formal education was also the one speaking three languages. Therefore a more direct measurement of literacy, such as the
North American Reading Test, might be helpful in interpreting the results. Overall, our group of patients was well functioning. Testing in a larger sample will be needed to assess whether the discriminative power of MicrocogTM is the same as in older non-cancer patients [13]. A major concern that prompted this feasibility study was to ensure that computer illiterate patients, or patients with visual or auditory impairments such as those encountered in our usual population of elderly cancer patients could complete the test easily. Most of our patients were computer illiterate, yet performed very well and enjoyed the test. Corrected visual impairments did not appear to interfere with the test. Similarly, the three patients with English as a second language scored similarly to the others. This test has some downsides. These are a somewhat difficult conversion into the SAS statistical software, which we corrected with a few lines of software, and the cost (from $ 6 to 10 per test, plus $ 250 for the program). However, one should note that paper tests cost about $ 5 per test. We believe that the gains in reproducibility, personnel needs, and patient acceptability make the investment worthwhile. In conclusion, MicrocogTM is a very feasible test in elderly cancer patients having received chemotherapy, with promising properties for studies in this patient population. This pilot also demonstrated in a more general way the feasibility of computerized cognitive testing in this population. Other computer tests of more modern design should be assayed as well, and their performance compared in this patient population. It is at this point too early to propose any direct clinical use of these tests in older cancer patients. Another challenge is the increasing number of patients with established dementia who may present a cancer. Little is known about treatment patterns and prognosis of these patients, even more about the cognitive impact of cancer treatment. MicrocogTM has been validated in dementia patients [13]. Its usefulness in cancer patients with established dementia remains to be evaluated. Reviewers Carol M. Moinpour, Ph.D., Behavioral Scientist/Southwest Oncology Group Statistical Center, Associate Member/Division of Public Health Sciences, Fred Hutchinson Cancer Research Center, 1100 Fairview Av. North/MP-557, Box 19024, Seattle, WA 98109-1024, USA. Pand´elis Giannakopoulos, M.D., FMH (Psy), Professor, Associate Vice-Dean, Service of Geriatric Psychiatry, Department of Psychiatry-HUG-Belle-Idee, Pavilion Les Champs, 2 Ch. du Petit-Bel-Air, CH-1225 Chˆene-Bourg, Geneva; Psychogeriatric Service, University of Lausanne, CH-1008 Prilly, Switzerland. Dr. Gilbert Zulian, Centre de Soins Continus (CESCO), Ch. de la Savonni`ere 11, CH-1245 Collonge-Bellerive, Switzerland.
M. Extermann et al. / Critical Reviews in Oncology/Hematology 54 (2005) 137–143
References [1] Ahles TA, Saykin A. Cognitive effects of standard-dose chemotherapy in patients with cancer. Cancer Invest 2001;19:812–20. [2] Schagen SB, van Dam FS, Muller MJ, Boogerd W, Lindeboom J, Bruning PF. Cognitive deficits after postoperative adjuvant chemotherapy for breast carcinoma. Cancer 1999;85:640– 50. [3] Cherrier MM, Rose AL, Higano C. The effects of combined androgen blockade on cognitive function during the first cycle of intermittent androgen suppression in patients with prostate cancer. J Urol 2003;170:1808–11. [4] Tchen N, Juffs HG, Downie FP, et al. Cognitive function, fatigue, and menopausal symptoms in women receiving adjuvant chemotherapy for breast cancer. J Clin Oncol 2003;21:4175–83. [5] Saykin AJ, Ahles TA, McDonald BC. Mechanisms of chemotherapyinduced cognitive disorders: neuropsychological, pathophysiological, and neuroimaging perspectives. Semin Clin Neuropsychiatry 2003;8:201–16. [6] Tannock IF, Ahles TA, Ganz PA, Van Dam FS. Cognitive impairment associated with chemotherapy for cancer: report of a workshop. J Clin Oncol 2004;22:2233–9. [7] Mattis S. Dementia rating scale. Odessa, FL: Psychological Assessment Resources; 1989. [8] Randolph C. RBANS manual—repeatable battery for the assessment of neuropsychological status. San Antonio, TX: The Psychological Corporation, Harcourt; 1998. [9] Powell DH, Kaplan EF, Whitla D, Weintraub S, Catlin R, Funkenstein HH. MicrocogTM manual. San Antonio, TX: The Psychological Corporation, Harcourt Brace & Co.; 1993. [10] Elwood RW. MicroCog: assessment of cognitive functioning. Neuropsychol Rev 2001;11:89–100. [11] Holliday SL, Costello RM. Screening for neuropsychological deficits in Gulf War veterans: MicroCog vs standard neuropsychological tests. Arch Clin Neuropsychol 1997;12:338. [12] Moore JL, McAuley JW, Long L, Bornstein R. An evaluation of the effects of methylphenidate on outcomes in adult epilepsy patients. Epilepsy Behav 2002;3:92–5. [13] Green RC, Green J, Harrison JM, Kutner MH. Screening for cognitive impairment in older individuals—validation study of a computerbased test. Arch Neurol 1994;51:779–86.
143
[14] Lopez SJ, Edwards LM, Floyd RK, Magyar-Moe J, Rehfeldt JD, Ryder JA. Note on comparability of Microcog test forms. Percept Motor Skills 2001;93:825–8. [15] Erlanger DM, Kaushik T, Broshek D, Freeman J, Feldman D, Festa J. Development and validation of a web-based screening tool for monitoring cognitive status. J Head Trauma Rehab 2002;17: 458–76. [16] Petrella JK, Miller LS, Cress ME. Leg extensor power, cognition, and functional performance in independent and marginally dependent older adults. Age and Ageing 2004;33:342–8. [17] Wesnes KA, McKeith IG, Ferrara R, et al. Effects of rivastigmine on cognitive function in dementia with lewy bodies: a randomised placebo-controlled international study using the cognitive drug research computerised assessment system. Dement Geriatr Cogn Disorder 2002;13:183–92.
Biographies Martine Extermann is Associate Professor in the Senior Adult Oncology Program (SAOP) at H. Lee Moffitt Cancer Center, University of South Florida (USF). Hongbin Chen was Assistant Director of the Florida Statistical Center on Aging at USF during the study. Margaret Booth-Jones is a neuropsychologist in the Psychosocial and Palliative Care Program at H. Lee Moffitt Cancer Center. Julie Meyer is the nurse practitioner of the SAOP. Lodovico Balducci is Professor and Program Leader of the SAOP. Paul Jacobsen is a psychologist and Program Leader of Program.