Lumbar puncture increases Alzheimer’s disease biomarker levels in cerebrospinal fluid of rhesus monkeys

Summary

Cerebrospinal fluid (CSF) samples are commonly collected via lumbar puncture (LP) in both clinical and research settings for measurement of biomarkers of Alzheimer’s disease (AD). To determine the effects of LP on CSF AD biomarkers, we collected CSF samples at seven different time points after an LP in rhesus monkeys. We find that amyloid-beta (Aβ) and Tau levels increased significantly on day 1, peaked on day 3, and returned to baseline on day 10 after LP. The NFL levels increased significantly on day 5, peaked on day 10, and returned to baseline after day 30. The increased AD biomarker levels were mainly due to CSF outflow and deep intrathecal invasion during LP. Therefore, if LPs are repeated within a short period of time, prior LP can affect Aβ and Tau levels within 10 days and NFL levels within 30 days, which may lead to clinical misdiagnosis or incorrect scientific conclusions.

Graphical abstract

Highlights

• •LP can lead to rapid and significant increase of CSF AD biomarker levels in monkeys
• •LP affects Aβ and Tau levels up to 10 days and NFL levels more than 30 days
• •The effects might be attributed to CSF outflow and deep intrathecal invasion
• •Clinical diagnosis or research results with repeated LP need to be cautious

Molecular medicine; Molecular neuroscience

Introduction

Alzheimer’s disease (AD) is the most common form of dementia worldwide, accounting for 60–80% of all dementia cases.¹ In 2018, Alzheimer’s Disease International (ADI) estimated that approximately 50 million people worldwide suffer from dementia, which is expected to grow to about 150 million by 2050,² making it one of the biggest global public health challenges. In 2018, the global cost of the disease was estimated at $1 trillion and is expected to increase to $2 trillion by 2030,² thus imposing a huge economic burden for society. Given the increasing prevalence and mortality of AD, as well as increasing health care costs, timely detection, accurate diagnosis, and development of effective treatments are urgently needed.

Specific biomarkers in cerebrospinal fluid (CSF) are crucial for AD diagnosis and are a direct reflection of typical pathologies, including amyloid-beta (Aβ) plaques, hyperphosphorylated Tau tangles, and neuronal degeneration in the brain.³ Among these biomarkers, Aβ42 levels and Aβ42/Aβ40 level ratios are negatively correlated with intracerebral Aβ load. Total Tau (t-Tau) and neurofilament light (NFL) levels directly reflect the extent of neuronal degeneration, and phosphorylated Tau (p-Tau) is considered a direct marker of tangle pathology.³ Comprehensive analysis of these biomarkers can provide an accurate prediction of AD,^4,5,6,7 forming the core of all current AD diagnostic criteria.^8,9,10,11,12 The National Institute on Aging and Alzheimer’s Association (NIA-AA) diagnostic criteria^8,9,10 are based on clinical symptoms as well as CSF and imaging biomarkers. The diagnostic criteria of the International Working Group (IWG)¹¹ are also based on CSF and imaging biomarkers. The NIA-AA research framework goes a step further, using biomarkers to define AD, considering cognitive impairment as a symptom of the disease rather than a definition.¹² Furthermore, regardless of AD diagnostic criteria, diagnosis in the preclinical asymptomatic stage is increasingly recognized as the optimal window for slowing or arresting AD progression. It relies almost exclusively on biomarkers, further emphasizing the importance of CSF AD biomarkers for early diagnosis and intervention.

CSF sample collection in clinical and research settings is routinely performed via lumbar puncture (LP), which makes LP an important tool in the clinical diagnosis and research of AD and other neurodegenerative diseases. For instance, in the Netherlands and Sweden, 40% of new AD patients were diagnosed by LP.¹³ In pharmacodynamic studies as well as in disease tracking, multiple LPs are often required for longitudinal comparisons of CSF AD biomarker levels.^14,15,16 Based on observations during the development of rhesus monkey AD models, we found that CSF AD biomarkers were abnormally elevated 1–2 days following the LP procedure. After a systematic literature search, we found only two studies reporting on this phenomenon^17,18, but with unclear and contradictory conclusions. Olsson et al. explored the relationship between sleep and AD biomarkers in humans and found that CSF Aβ and Tau levels were elevated in two out of four subjects 3 days after LP, whereas NFL levels were unchanged.¹⁷ In contrast, Boehnke et al. reported that NFL levels were significantly elevated in the CSF of macaque monkeys 14–23 days after LP, while Aβ and Tau levels were unchanged.¹⁸ Furthermore, in human-based studies where two LPs were conducted within a short period, no changes in AD biomarker levels were detected, although this finding may have been influenced by other events occurring between the two LPs.^19,20 The current paucity of research in this field demands immediate attention, given the routine use of LP and its potential impact on CSF AD biomarkers and accuracy of subsequent AD diagnosis. In research settings where longitudinal comparisons are often required,²¹ such as studies of diurnal CSF protein variation^22,23, the influence of multiple LPs may lead to erroneous findings and misleading conclusions. Thus, rigorous and comprehensive studies investigating the effects of LP on CSF AD biomarkers are urgently needed.

In the present study, we systematically investigated the effect of LP on CSF AD biomarkers to address the above questions. First, we characterized the duration of the effects of LP on Aβ, Tau, and NFL levels to determine the extent of their impact on AD diagnosis. Second, we investigated which of the three main factors involved in LP, i.e., CSF outflow, deep intrathecal invasion, and anesthesia, led to changes in AD biomarker levels, thereby providing guidance for the proper application of LP in future diagnosis and study.

Results

Time-course study: Temporal effects of LP on CSF AD biomarker levels

Fifty-six monkeys were randomly divided into seven groups. Each monkey underwent an initial LP (LP1), during which CSF samples were collected as the baseline. A second LP (LP2) was performed on each monkey at 1, 3, 5, 7, 10, 30, or 100 days after LP1, depending on group assignment (see method details). During the period between the two LPs, the monkeys were fed a normal diet and did not undergo any other operative procedures. Therefore, any changes in CSF levels of the AD biomarkers observed between LP1 and LP2 could be attributed to the initial LP. Details on the effects of LP on Aβ, Tau, and NFL levels are described below.

LP caused a rapid increase and fast return to baseline in CSF Aβ levels

To investigate the effects of LP on CSF Aβ levels, the percentage change in Aβ levels from LP1 to LP2 was calculated. Analysis revealed that the effects of LP on CSF Aβ lasted approximately 10 days. Therefore, the levels of Aβ40 and Aβ42 were only examined in the 1-day, 3-day, 5-day, 7-day, and 10-day groups.

As shown in Figures 1A–1E, the Aβ40 levels in LP2 were significantly higher than those in LP1 in the 1-, 3-, and 5-day groups, but not significantly different from baseline in the 7- and 10-day groups. As shown in Figures 1F–1J, the Aβ42 levels in LP2 were significantly higher than those in LP1 in the 1-, 3-, 5-, and 7-day groups, but not significantly different from baseline in the 10-day groups.

Figure 1.

Temporal effects of LP on CSF Aβ levels

Monkeys in five groups were subjected to LP1, with collected CSF samples used as baseline. LP2 was subsequently performed to collect CSF after 1 (1-day group), 3 (3-day group), 5 (5-day group), 7 (7-day group), and 10 days (10-day group), respectively. The LP1 and LP2 levels of Aβ40 and Aβ42 in each group were measured.

(A–E) Aβ40 levels were significantly higher in LP2 than in LP1 in 1-day (mean: LP1 = 1 997 pg/mL, LP2 = 3 696 pg/mL, paired t-test, p = 0.0019), 3-day (mean: LP1 = 2 031 pg/mL, LP2 = 5 677 pg/mL, paired t-test, p = 0.0003), and 5-day groups (mean: LP1 = 2 156 pg/mL, LP2 = 4 121 pg/mL, paired t-test, p = 0.0186), but not significantly different from baseline in the 7-day and 10-day groups.

(F–J) Aβ42 levels were significantly higher in LP2 than in LP1 in the 1-day (mean: LP1 = 783.5 pg/mL, LP2 = 1361 pg/mL, paired t-test, p = 0.0039), 3-day (mean: LP1 = 765.0 pg/mL, LP2 = 2 592 pg/mL, paired t-test, p = 0.0007), 5-day (mean: LP1 = 802.3 pg/mL, LP2 = 1 680 pg/mL, paired t-test, p = 0.0297), and 7-day groups (mean: LP1 = 1 049 pg/mL, LP2 = 2 042 pg/mL, Wilcoxon signed-rank test, p = 0.0391), but not significantly different from baseline in the 10-day group.

(K and L) Mean ± SEM of the percentage change in Aβ40 and Aβ42 levels from LP1 to LP2 (% change = [LP2 − LP1]/LP1 × 100) is plotted against LP2 sampling time. Horizontal axis represents time interval between LP1 and LP2 for different groups. Levels of Aβ40 and Aβ42 after LP increased significantly on day 1 (mean % change = 86% for Aβ40, 84% for Aβ42), peaked on day 3 (mean % change = 203% for Aβ40, 303% for Aβ42), decreased rapidly on days 5 (mean % change = 88% for Aβ40, 117% for Aβ42) and 7 (mean % change = 74% for Aβ40, 89% for Aβ42), and returned to baseline on day 10 (mean % change = 3% for Aβ40, 12% for Aβ42). Error bars indicate mean ± SEM. ns: p > 0.05, ∗: p < 0.05, ∗∗: p < 0.01, ∗∗∗: p < 0.001; Aβ: Amyloid-beta; LP: Lumbar puncture. See also Figure S1.

Figures 1K–1L shows the mean and standard error of the mean (mean ± SEM) of the percentage change in Aβ levels (% change = [LP2 – LP1]/LP1 × 100) for all the monkeys in each group against LP2 sampling time. As shown in Figures 1K and 1L, LP1 sampling resulted in a rapid and significant increase in the CSF Aβ40 and Aβ42 levels on day 1, which peaked on day 3, with a mean percentage change of 203% for Aβ40 of 303% for Aβ42. The levels then decreased rapidly on days 5 and 7 before returning to baseline on day 10.

LP caused a rapid increase and fast return to baseline in CSF Tau levels

The same strategy was used to investigate the effects of LP on CSF Tau levels. As shown in Figures 2A–2O, t-Tau levels were significantly higher in LP2 than in LP1 in the 1-, 3-, 5, 7-day groups, but not significantly different from baseline in the 10-day group. Furthermore, the p-Tau181 and p-Tau231 levels were significantly higher in LP2 than in LP1 in the 1- and 3-day groups, but not significantly different from baseline in the 5-, 7-, and 10-day groups.

Figure 2.

Temporal effects of LP on CSF Tau levels

Monkeys in five groups were subjected to LP1, with collected CSF samples used as baseline. LP2 was subsequently performed to collect CSF after 1 (1-day group), 3 (3-day group), 5 (5-day group), 7 (7-day group), and 10 days (10-day group), respectively. The LP1 and LP2 levels of t-Tau, p-Tau181, and p-Tau231 in each group were measured.

(A–E) Levels of t-Tau were significantly higher in LP2 than in LP1 in 1-day (mean: LP1 = 153.7 pg/mL, LP2 = 426.3 pg/mL, paired t-test, p = 0.0131), 3-day (mean: LP1 = 105.7 pg/mL, LP2 = 567.0 pg/mL, paired t-test, p = 0.0024), 5-day (mean: LP1 = 121.5 pg/mL, LP2 = 352.7 pg/mL, paired t-test, p = 0.0449), and 7-day groups (mean: LP1 = 138.7 pg/mL, LP2 = 328.1 pg/mL, Wilcoxon signed-rank test, p = 0.0391), but not in 10-day group.

(F–J) Levels of p-Tau181 were significantly higher in LP2 than in LP1 in 1-day (mean: LP1 = 68.57 pg/mL, LP2 = 115.4 pg/mL, paired t-test, p = 0.0333) and 3-day groups (mean: LP1 = 36.86 pg/mL, LP2 = 101.4 pg/mL, Mann-Whitney U test, p = 0.0002), but not in 5-day, 7-day, and 10-day groups.

(K–O) Levels of p-Tau231 were significantly higher in LP2 than in LP1 in 1-day (mean: LP1 = 67.11 pg/mL, LP2 = 107.2 pg/mL, Wilcoxon signed-rank test, p = 0.0078) and 3-day groups (mean: LP1 = 35.70 pg/mL, LP2 = 102.5 pg/mL, paired t-test, p < 0.0001), but not in 5-day, 7-day, and 10-day groups.

(P–R) Mean ± SEM of the percentage change in Tau levels from LP1 to LP2 is plotted against LP2 sampling time. Horizontal axis represents time interval between LP1 and LP2 for different groups. Levels of t-Tau, p-Tau181, and p-Tau231 after LP increased significantly on day 1 (mean % change = 188% for t-Tau, 82% for p-Tau181, 73% for p-Tau231), peaked on day 3 (mean % change = 455% for t-Tau, 204% for p-Tau181, 204% for p-Tau231), decreased rapidly on days 5 (mean % change = 203% for t-Tau, 86% for p-Tau181, 53% for p-Tau231) and 7 (mean % change = 141% for t-Tau, 28% for p-Tau181, 50% for p-Tau231), and returned to baseline on day 10 (mean % change = 6% for t-Tau, −8% for p-Tau181, −7% for p-Tau231). Error bars indicate mean ± SEM. ns: p > 0.05, ∗: p < 0.05, ∗∗: p < 0.01, ∗∗∗∗: p < 0.0001; LP: Lumbar puncture. See also Figure S1.

Figure 2P-R shows the mean ± SEM of the percentage change in CSF Tau levels from LP1 to LP2 for all the monkeys in each group against LP2 sampling time. As seen in Figures 2P–2R, the levels of t-Tau, p-Tau181, and p-Tau231 after LP increased significantly on day 1, peaked on day 3 (mean % change = 455% for t-Tau, 204% for p-Tau181, and 204% for p-Tau231), decreased rapidly on days 5 and 7, and returned to baseline on day 10.

LP caused a slow long-term increase and slow return to baseline in CSF NFL levels

The CSF levels of NFL in all seven groups (1-day, 3-day, 5-day, 7-day, 10-day, 30-day, and 100-day groups) were examined. As shown in Figures 3A–3G, NFL levels were significantly higher in LP2 than in LP1 in the 5-, 7-, 10, and 30-day groups, but not significantly different from baseline in the 1-, 3-, and 100-day groups. Figure 3H shows the mean ± SEM of the percentage change in CSF NFL levels from LP1 to LP2 for each monkey in each group against LP sampling time. As seen in Figure 3H, NFL levels after LP were comparable to baseline on day 1 (mean % change = −6%), showed a non-significant upward trend on days 3 (mean % change = 68%), increased significantly on days 5 (mean % change = 156%) and 7 (mean % change = 295%), peaked on day 10 (mean % change = 687%), decreased substantially on day 30 (mean % change = 195%), and returned to baseline on day 100 (mean % change = −19%).

Figure 3.

Temporal effects of LP on CSF NFL levels

Monkeys in seven groups were subjected to LP1, with collected CSF samples used as baseline. LP2 was subsequently performed to collect CSF after 1 (1-day group), 3 (3-day group), 5 (5-day group), 7 (7-day group), 10 (10-day group), 30 (30-day group), and 100 days (100-day group), respectively. The LP1 and LP2 levels of NFL in each group were measured.

(A–G) NFL levels were significantly higher in LP2 than in LP1 in 5-day (mean: LP1 = 502.7 pg/mL, LP2 = 1 372 pg/mL, Wilcoxon signed-rank test, p = 0.0469), 7-day (mean: LP1 = 561.4 pg/mL, LP2 = 2269 pg/mL, Wilcoxon signed-rank test, p = 0.0156), 10-day (mean: LP1 = 501.0 pg/mL, LP2 = 3112 pg/mL, paired t-test, p = 0.0175), and 30-day groups (mean: LP1 = 596.0 pg/mL, LP2 = 1 542 pg/mL, Wilcoxon signed-rank test, p = 0.0391), but not in 1-day, 3-day, and 100-day groups.

(H) Mean ± SEM of the percentage change in NFL levels from LP1 to LP2 is plotted against LP2 sampling time. Horizontal axis represents time interval between LP1 and LP2 for different groups. Levels of NFL after LP were comparable to baseline on day 1 (mean % change = −6%), showed a non-significant upward trend on day 3 (mean % change = 68%), increased statistically on days 5 (mean % change = 156%), and day 7 (mean % change = 295%), peaked on day 10 (mean % change = 687%), decreased substantially on day 30 (mean % change = 195%), and returned to baseline before day 100 (mean % change = −19%). Error bars indicate mean ± SEM. ns: p > 0.05, ∗: p < 0.05; NFL: Neurofilament light; LP: Lumbar puncture. See also Figure S1.

Mechanistic study: Causes of increased levels of CSF AD biomarkers

After comprehensive analysis of the LP process and relevant literature, we identified three possible factors that may influence AD biomarker levels during LP: 1) CSF outflow during LP sampling^24,25; 2) depth of intrathecal invasion during LP^26,27,28; and 3) anesthesia administered to the monkeys prior to LP.^29,30 In the time-course study, we demonstrated that the temporal changing pattern of NFL levels after LP differed from that of Aβ and Tau. To investigate the potential effects of these three factors on AD biomarker levels, we established six additional monkey groups, with three dedicated to studying the effects on Aβ and Tau and the remaining three dedicated to studying the effects on NFL.

CSF outflow may be the main cause for the increase in Aβ levels

To investigate the effects of the three above factors on Aβ levels in LP, we conducted three separate experiments on three groups of monkeys in Set A. In the first group (“LP with minimal CSF-outflow” (LP-MO) group in Set A), we studied CSF outflow; in the second group (“LP with minimal intrathecal invasion” (LP-MI group in Set A), we studied deep intrathecal invasion, and in the third group (anesthesia group in Set A), we studied anesthesia. The three groups were subjected to two special LPs (LP-MO, LP-MI) and anesthesia-only procedure, respectively, more than 10 days after baseline sample collection by LP1. Second CSF samples were collected by LP2 three days after the LP-MO, LP-MI and anesthesia-only procedure. For the LP-MO and LP-MI groups, we compared the levels of Aβ40 and Aβ42 in LP2 with those of the baseline samples (LP1) and obtained peak percentage changes in CSF Aβ40 and Aβ42 after LP-MO and LP-MI relative to baseline. If differences in peak percentage changes were observed after LP-MO or LP-MI compared with those after normal LP1 (3-day group), it would suggest that CSF outflow or deep intrathecal invasion had an effect on CSF Aβ levels. If differences in peak percentage changes were not observed after LP-MO or LP-MI compared with those after normal LP1 (3-day group), it would suggest that CSF outflow or deep intrathecal invasion had no effect on CSF Aβ levels. For the anesthesia factor study, if differences in Aβ40 and Aβ42 levels between LP2 and LP1 were observed, it would indicate that anesthesia had an effect on CSF Aβ levels, as only the anesthesia procedure occurred between the two samplings.

In the LP-MO group in Set A, significant differences were observed in the peak percentage changes in CSF Aβ40 and Aβ42 after LP-MO compared to that after normal LP1 (Figures 4A and 4B). This suggests that CSF outflow in normal LP can have a significant effect on CSF Aβ40 and Aβ42 levels. In the LP-MI group in Set A, no significant differences were observed in the peak percentage changes in CSF Aβ40 and Aβ42 levels after LP-MI compared to that after normal LP1 (Figures 4C and 4D). This suggests that deep intrathecal invasion does not affect CSF Aβ40 and Aβ42 levels. Figure S5 provides detailed data and statistical analysis results of Aβ40 and Aβ42 levels in each monkey in the LP-MO and LP-MI groups in Set A. In the anesthesia group in Set A (Figures 4E and 4F), no significant changes were observed in the levels of Aβ40 and Aβ42 in LP2 compared to LP1, despite an anesthesia-only procedure being applied between the LP samplings. This suggests that anesthesia does not cause significant changes in Aβ levels.

Figure 4.

Effects of CSF outflow, intrathecal invasion, and anesthesia on Aβ40 and Aβ42 levels

More than 10 days after baseline sample collection (LP1), three groups of monkeys were subjected to LP-MO to determine effect of CSF outflow (LP-MO group in Set A), LP-MI to determine effect of deep intrathecal invasion (LP-MI group in Set A), and anesthesia-only procedure to determine effect of anesthetic (anesthesia group in Set A). LP2 CSF samples were collected three days after above three operations, and LP2 levels of Aβ40 and Aβ42 were compared with those at baseline (LP1) for each group. Peak percentage changes in CSF Aβ40 and Aβ42 levels after LP-MO and LP-MI relative to baseline were obtained.

(A and B) Peak percentage changes in CSF Aβ40 (Mann-Whitney U test, p = 0.0003) and Aβ42 levels (Mann-Whitney U test, p = 0.0003) after LP-MO differed significantly from those after normal LP1.

(C and D) Peak percentage changes in CSF Aβ40 and Aβ42 levels after LP-MI did not differ significantly from those after normal LP1.

(E and F) There were no significant changes in Aβ40 and Aβ42 levels in LP2 compared to LP1. Taken together, CSF outflow may be the main cause for the increase in Aβ levels after LP sampling. Error bars indicate mean ± SEM. ns: p > 0.05, ∗∗∗: p < 0.001. Aβ: Amyloid-beta; LP: Lumbar puncture; LP-MO: LP without CSF-outflow; LP-MI: LP without intrathecal invasion. See also Figures S3–S5.

Thus, taken together, CSF outflow may be the main cause for the increase in Aβ levels after LP sampling.

Deep intrathecal invasion and CSF outflow may be the main causes for the increase in Tau levels

We also compared the levels of t-Tau, p-Tau181, and p-Tau231 in LP2 versus LP1 in the LP-MO group, LP-MI group, and anesthesia group in Set A, which were just discussed above.

In the LP-MO group in Set A, peak percentage changes in CSF t-Tau, p-Tau181 and p-Tau231 levels after LP-MO differed significantly from those after normal LP1 in time-course study (Figures 5A–5C). This suggests that CSF outflow in normal LP had a significant effect on CSF t-Tau, p-Tau181, and p-Tau231 levels. In the LP-MI group in Set A, peak percentage changes in CSF t-Tau, p-Tau181, and p-Tau231 levels after LP-MI also differed significantly from those after normal LP1 (Figures 5D–5F). This suggests that deep intrathecal invasion had a significant effect on CSF t-Tau, p-Tau181, and p-Tau231 levels. Figure S6 provides detailed data and statistical analysis results of t-Tau, p-Tau181, and p-Tau231 of each monkey in the LP-MO and LP-MI groups in Set A. In the anesthesia group in Set A (Figures 5G–5I), no significant changes were observed in the levels of t-Tau, p-Tau181, and p-Tau231 in LP2 compared to LP1, despite an anesthesia-only procedure being applied between the LP samplings. This suggests that anesthesia did not have a significant effect on t-Tau, p-Tau181, and p-Tau231 levels.

Figure 5.

Effects of CSF outflow, intrathecal invasion, and anesthesia on t-Tau, p-Tau181, and p-Tau231 levels

More than 10 days after baseline sample collection (LP1), three groups of monkeys in Set A were subjected to LP-MO to determine effect of CSF outflow (LP-MO group in Set A), LP-MI to determine effect of intrathecal invasion (LP-MI group in Set A), and anesthesia-only procedure to determine effect of anesthetic (anesthesia group in Set A). LP2 CSF samples were collected three days after the three operations, and LP2 levels of t-Tau, p-Tau181 and p-Tau231 were compared with those at baseline (LP1) for each group. Peak percentage changes in CSF t-Tau, p-Tau181, and p-Tau231 after LP-MO and LP-MI relative to baseline were obtained.

(A–C) Peak percentage changes in CSF t-Tau (Mann-Whitney U test, p = 0.0007), p-Tau181 (Mann-Whitney U test, p = 0.0006), and p-Tau231 levels (Mann-Whitney U test, p = 0.0007) after LP-MO differed significantly from those after normal LP1.

(D–F) Peak percentage changes in CSF t-Tau (Mann-Whitney U test, p = 0.0002), p-Tau181 (Mann-Whitney U test, p = 0.0002), and p-Tau231 levels (Unpaired t-test, p = 0.0007) after LP-MI differed significantly from those after normal LP1.

(G–I) There were no significant changes in t-Tau, p-Tau181, and p-Tau231 levels in LP2 compared to LP1. Taken together, both intrathecal invasion factor and CSF outflow factor may be the primary factors affecting Tau levels after LP sampling. Error bars indicate mean ± SEM. ns: p > 0.05, ∗∗∗: p < 0.001. LP: Lumbar puncture; LP-MO: LP without CSF-outflow; LP-MI: LP without intrathecal invasion. See also Figures S3, S4, and S6.

Thus, taken together, both deep intrathecal invasion factor and CSF outflow factor may be the primary factors affecting Tau levels after LP sampling.

Deep intrathecal invasion might be the cause of increase in NFL levels

To investigate the effects of the three factors on elevated NFL levels, we conducted three separate experiments on monkeys in Set B. In the first group (LP-MO group in Set B), we studied the CSF outflow factor; in the second group (LP-MI group in Set B), we studied the deep intrathecal invasion factor, and in the third group (anesthesia group in Set B), we studied the anesthetic factor. The CSF samples obtained during LP-MO and LP-MI were used as the baseline samples (LP1) for the LP-MO and LP-MI groups in Set B, respectively. The baseline sample for the anesthesia group in Set B was the LP1 CSF sample collected more than 100 days earlier. Second CSF samples (LP2) were collected 10 days after the operations. For the LP-MO and LP-MI groups, we compared the levels of NFL in LP2 with those at baseline (LP1) for each group (see method details) and obtained peak percentage changes in CSF NFL levels after LP-MO and LP-MI relative to baseline. If differences in peak percentage changes were observed after LP-MO or LP-MI compared with those after normal LP1 (10-day group), it would indicate that CSF outflow or deep intrathecal invasion had an effect on CSF NFL levels. If differences in peak percentage changes were not observed after LP-MO or LP-MI compared with those after normal LP1 (10-day group), it would indicate that CSF outflow or deep intrathecal invasion had no effect on CSF NFL levels. For the anesthesia factor study, if differences in NFL levels between LP2 and LP1 were observed, it would indicate that anesthesia had an effect on CSF NFL levels, as only the anesthesia procedure occurred between the two samplings.

In the LP-MO group in Set B, the peak percentage change in CSF NFL levels after LP-MO did not differ significantly from those after normal LP1 in time-course study (Figure 6D), indicating that CSF outflow in normal LP did not affect CSF NFL levels. In the LP-MI group in Set B, the peak percentage change in CSF NFL levels after LP-MI differed significantly from those after normal LP1 (Figure 6E), indicating that deep intrathecal invasion had a significant effect on CSF NFL levels. Figure S7 provides detailed data and statistical analysis results of NFL levels of each monkey in the LP-MO and LP-MI groups in Set B. In the anesthesia group in Set B (Figure 6F), no significant changes were observed in the levels of NFL in LP2 compared to LP1, despite an anesthesia-only procedure being applied between the LP samplings, suggesting that anesthesia did not have a significant effect on changes in NFL levels.

Figure 6.

Effect of CSF outflow, intrathecal invasion, and anesthesia on NFL levels

Three groups of monkeys in Set B were subjected to LP-MO to determine effect of CSF outflow (A, LP-MO group in Set B), LP-MI to determine effect of deep intrathecal invasion (B, LP-MI group in Set B), and anesthesia-only procedure to determine effect of anesthetic (C, anesthesia group in Set B). LP2 CSF samples were collected 10 days after the operations, and LP2 NFL levels were compared with those at baseline (LP1) for each group. Peak percentage changes in CSF NFL levels after LP-MO and LP-MI relative to baseline were obtained.

(D) Peak percentage changes in CSF NFL levels after LP-MO did not differ significantly from those after normal LP1.

(E) Peak percentage changes in CSF NFL levels after LP-MI differed significantly from those after normal LP1 (Mann-Whitney U test, p = 0.0111).

(F) No significant changes were observed in NFL levels in LP2 relative to LP1. Taken together, deep intrathecal invasion might be the main cause for the increase in NFL levels after LP sampling. Error bars indicate mean ± SEM. ns: p > 0.05, ∗: p < 0.05. NFL: Neurofilament light; LP: Lumbar puncture; LP-MO: LP without CSF-outflow; LP-MI: LP without intrathecal invasion. See also Figure S7.

Thus, taken together, deep intrathecal invasion might be the cause for the increase in NFL levels after LP sampling.

Discussion

In time-course study, we observed significant increases in Aβ and Tau levels one day after LP (mean increase of 73–188% across the biomarkers), although the precise onset of this increase remains unclear and warrants further investigation.

The data additionally revealed that LP markedly impacted the levels of Aβ and Tau within 10 days of the procedure, corroborating previous findings by Olsson et al. in humans.¹⁷ This similarity lends credence to the translation of results from studies involving monkeys to human contexts. LP also profoundly affected levels of NFL within 30 days post-LP, mirroring prior findings in a monkey cohort. Notably, Boehnke et al.¹⁸ documented a 200% increase in CSF NFL levels in monkeys 14–23 days after LP, while the present study observed an even more pronounced increase of approximately 400% when extrapolated to the same time frame. Thus, in clinical diagnostic or research settings where repeated LPs are performed within a short period of time, the previous LP can interfere with the biomarker levels of the next LP, leading to clinical misdiagnosis or incorrect scientific findings. Specifically, for Aβ and Tau, repeat LPs performed within 10 days could affect the results of the analysis, and for NFL, repeat LPs performed within 30 days could affect the results of the analysis. These findings challenge many prior studies using continuous lumbar drain or repeated LPs within 24 h to demonstrate diurnal variations in CSF proteins, suggesting that the proposed diurnal variation may be entirely due to changes in CSF flow.^22,23 Furthermore, these results could necessitate a revision of current practices related to the timing and methodology for repeat LPs, especially in instances where the initial attempt yields insufficient CSF. Therefore, we would alert the AD clinical and research fields to the significant impact of repeated LPs on CSF biomarker levels. When analyzing CSF biomarkers for AD diagnosis, investigations should be made to determine whether patients have a recent history of LP. Interference between LPs should be taken into account when designing studies with repeat LPs, especially in the fields of disease tracking and pharmacodynamic studies where multiple LPs are required.^14,15,16 Meanwhile, the conclusions of studies where multiple LPs exist should be treated with caution. To ensure the accuracy of clinical diagnosis and research results, we recommend that CSF sampling for Aβ and Tau should be conducted at least 10 days after a prior LP, while NFL testing should be performed more than 30 days after.

The Mechanistic Study revealed that the increase in Aβ was primarily due to CSF outflow, the increase in Tau was influenced by both CSF outflow and deep intrathecal invasion, and the elevation in NFL may be due to deep intrathecal invasion, rather than CSF outflow. This conclusion for NFL is corroborated by the observation of Boehnke et al. that NFL levels are not increased after cisterna magna puncture, despite similar CSF outflow.¹⁸ Consequently, LP procedures should aim to minimize both CSF outflow and intrathecal invasion to avoid these effects on AD biomarkers. To minimize the effects of LP on Aβ and Tau levels, the volume of CSF outflow should be minimized during sample collection and atraumatic spinal and narrow-bore needles should be used to reduce the risk of CSF leakage into the epidural space and thus CSF outflow.^31,32 Additionally, the use of ultrasound or alternative imaging techniques is recommended to guide LP and limit intrathecal invasion, thereby decreasing the likelihood of elevated Tau and NFL levels. In studies utilizing CSF Tau or NFL as biomarkers, cisterna magna puncture may also be an alternative to consider. By reducing CSF outflow volume and minimizing intrathecal invasion, it should be possible to minimize the effects of LP on AD biomarkers.

Different kinds of CSF leakage due to head injury, iatrogenic injury, and intracranial tumor,^33,34,35,36 as well as intrathecal invasion due to lumbar trauma, and spinal anesthesia,³⁷ are commonly seen in clinical practice. According to our results, both deep intrathecal invasion and CSF outflow can lead to elevated CSF AD biomarker levels. Thus, it is critical to confirm the presence of deep intrathecal invasion or CSF leakage from other causes before interpreting CSF AD biomarker test results for the diagnosis of AD, even if the patient has no prior history of LP.

The findings of our study suggest that the elevation in CSF NFL levels could be attributed to deep intrathecal invasion. In alignment with our conclusions, Boehnke et al. proposed that the increase in NFL was a consequence of cauda equina damage rather than CSF outflow, a hypothesis substantiated by their observation that repeated cisterna magna puncture did not result in increased levels of NFL.

Our results indicate that deep intrathecal invasion and CSF outflow may be the main causes for the increase in Tau levels. We speculate that deep intrathecal invasion may lead to nerve irritation, subsequently stimulating the release of Tau from the Tau-rich cauda equina located in the lumbar region.²⁷ In addition, the reduction in CSF volume caused by CSF outflow leads to a reduction in intracranial pressure.^38,39,40,41 This reduction affects CSF circulation and reduces CSF turnover,⁴² thereby limiting Tau drainage from the central to peripheral circulatory system, resulting in the accumulation of Tau in the CSF.²⁵ The interplay of these two factors is necessary for Tau accumulation in the CSF. Under effective Tau drainage, Tau released into the CSF as a result of intrathecal invasion is drained into the peripheral system, thereby maintaining normal CSF Tau levels. This phenomenon likely explains why Tau concentrations remained unchanged after the LP-MO experiment. Conversely, when intrathecal invasion does not occur, there is no supplementary release of Tau into the CSF, aligning with the findings of stable Tau concentrations after the LP-MI procedure.

Studies have suggested that CSF production is reduced in aging people.⁴³ This reduction is hypothesized to cause a reduction in CSF turnover, which reduces Aβ and Tau clearance from the brain to peripheral blood, causing an accumulation of both biomarkers in the brain and ultimately inducing AD onset.^24,25 In this study, CSF outflow caused by LP sampling led to the increase of Aβ and Tau in the CSF, suggesting that CSF volume reduction probably can lead to Aβ and Tau accumulation in the brain, thus providing additional evidence for the above hypothesis regarding AD pathogenesis.

To date, only two studies have explored the effects of LP on CSF AD biomarkers. The first study investigated the relationship between sleep and AD biomarkers in humans, reporting that CSF Aβ and Tau levels were elevated in two out of four subjects 3 days after LP, while NFL levels were unaltered.¹⁷ According to our data, the absence of NFL level changes in this previous study may be due to the short interval between LPs, with NFL levels not yet increasing at the time of the second LP. Moreover, this previous study failed to provide a clear conclusion regarding whether LP can cause elevated Aβ and Tau levels due to the small subject size and contradictory results among subjects. In the current study, we addressed these limitations and clearly showed that LP causes elevated Aβ and Tau levels. In the second study, which was conducted in macaque monkeys, results showed that CSF NFL levels were significantly increased 14–23 days after LP, while Aβ and Tau levels were unaltered.¹⁸ According to our data, the absence of changes in Aβ and Tau levels in the previous study may be due to the long interval between the two LPs such that the levels of Aβ and Tau had returned to baseline by the second LP sampling.

In conclusion, the present study demonstrates that the LP procedure, a primary clinical method for CSF collection, can significantly increase CSF Aβ and Tau levels in a short time period. Using a multiple time point sampling method, we were able to map the course of AD biomarker levels over time. As rhesus monkeys are phylogenetically close to humans,⁴⁴ with similar anatomical structures and functions, and are subjected to identical LP procedures,⁴⁵ we believe that our experimental results can probably be extrapolated to humans. If our results are confirmed in patients, they should be taken into account by clinicians and AD researchers applying LPs at multiple time points. Additionally, as Tau and NFL are used as biomarkers in a variety of neurodegenerative diseases (e.g., multiple sclerosis, frontotemporal dementia, stroke), this result is of relevance to the broader neurodegeneration community beyond those studying AD. Based on our data, we recommend minimizing the volume of sample taken and avoiding multiple LPs during a short time frame. We propose several modifications to reduce the effects of LP on AD biomarkers, including minimizing CSF outflow volume during collection, using atraumatic spinal and narrow-bore needles to reduce the risk of CSF leakage into the epidural space and thus CSF loss, and employing ultrasound imaging or other visualization methods to guide LP needle penetration to minimize intrathecal invasion.

Limitations of the study

In the time-course study, CSF Aβ and Tau levels in monkeys have already significantly elevated after one day of LP, and the exact onset time of the elevation, which is critical for the clinical application of LP, is still unclear and needs further investigation. In the Mechanistic study, we only investigated which factors in LP led to the elevation of CSF AD biomarkers. Further research investigating the specific mechanisms is needed. In addition, the present study was conducted on rhesus monkeys. Although monkeys and humans are phylogenetically similar, resulting in highly similar structures and functions between their central nervous systems, it is still necessary to further validate our results on humans.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Biological samples
Monkey cerebrospinal fluid samples	This study	N/A
Critical commercial assays
Human β-Amyloid (1–40) ELISA kit	Thermo Fisher	Cat# KHB3481; RRID: NA
β-Amyloid (1–42) ELISA kits	Thermo Fisher	Cat# KHB3441; RRID: NA
Human Tau assay kit	Quanterix	Cat# 101552; RRID: NA
Human P-Tau 181 V2 assay kit	Quanterix	Cat# 103714; RRID: NA
Human P-Tau 231 assay kit	Quanterix	Cat# 102292; RRID: NA
Human NFL assay kit	Quanterix	Cat# 103186; RRID: NA
Software and algorithms
Prism statistics	GraphPad	Version 8.0.2; RRID: SCR_002798
Other
Spinal needle	Qionghua	Cat# GCZ-10; RRID: NA
Ultrasonic imaging system	ESAOTA	PML# MYLAB 70VET XV; RRID: NA
Puncture rack	Jingfang	Cat# JSM-113; RRID: NA

Resource availability

Lead contact

Further information and requests for resources should be directed to the lead contact, Xintian Hu (xthu@mail.kiz.ac.cn).

Materials availability

This study did not generate new unique reagents.

Data and code availability

All data will be shared upon request to the lead contact. No standardized datatype data were generated in this study. This study did not generate new code. Any additional analysis information for this work is available by request to the lead contact.

Experimental model and study participant details

Animals

Seventy-seven healthy male rhesus macaques (Macaca mulatta, 4–17 years, 5.0–13.6 kg, Table S1) purchased from the Kunming Primate Research Center (Yunnan, China) were used in this study. Of these, 34 were housed in large colonies (5–16 monkeys in each colony) under natural light, with connected indoor (2.65 × 2.40 × 2.86 m) and outdoor (3.52 × 2.64 × 3.13 m) areas. The remaining 43 monkeys were housed in single cages (0.80 × 0.80 × 0.80 m) in rooms under standard experimental conditions (i.e., 22 ± 1°C temperature, 50% ± 5% relative humidity, and 12 h light/12 h dark cycle (light on at 07:00 am)). All monkeys were provided with monkey biscuits twice a day, fruit and vegetables once a day, and water ad libitum. Professional veterinarians provided daily care throughout the experiment. All experimental procedures were performed in strict compliance with the National Institutes of Health (USA) Guide for the Care and Use of Laboratory Animals (8th edition, 2011; National Academies Press, Washington DC, USA) and the Regulations on the Health and Management of Laboratory Animals of the Kunming Institute of Zoology, Chinese Academy of Sciences. All experimental protocols were approved by the Ethics Committee of the Kunming Primate Research Center (Approval ID: IACUC-PE-2021-12-001).

Method details

Experimental design

Time-course study: Temporal effects of LP on CSF AD biomarker levels

Fifty-six monkeys were randomly assigned to seven groups. An initial LP (LP1) was performed on each monkey, with the resulting CSF samples used as the baseline. A second LP (LP2) was performed on each monkey in all seven groups to collect CSF samples after 1 day (1-day group, n = 8), 3 days (3-day group, n = 8), 5 days (5-day group, n = 8), 7 days (7-day group, n = 8), 10 days (10-day group, n = 8), 30 days (30-day group, n = 8), and 100 days (100-day group, n = 8) (Figure S1). Changes in the CSF levels of Aβ, Tau, and NFL from the LP2 samples relative to the baseline LP1 samples were determined to measure the effects of LP1 on Aβ, Tau, and NFL levels at different time points. In this study, we determined the onset and duration of the effects of LP on CSF Aβ, Tau, and NFL levels.

Mechanistic study: Potential causes of changes in CSF AD biomarker levels after LP

Another cohort of monkeys (n = 45) was used to investigate the causes of changes in CSF AD biomarker levels after LP. Based on literature review and analysis of LP procedures, we identified three potential factors as primary causes.

Effect of CSF outflow during LP on changes in AD biomarkers

During LP sampling in time-course study, 1.5 mL of CSF per monkey was collected, accounting for approximately 10% of the total CSF volume in monkeys^21,46 (Figure S2A). This non-negligible CSF outflow can decrease intracranial pressure and consequently decrease the rate of CSF circulation, ultimately reducing drainage of AD biomarkers from the central to peripheral circulatory system and leading to an increase in CSF AD biomarker levels.^24,25

To explore the potential role of CSF outflow on changes in AD biomarker levels, an LP with minimal CSF-outflow (LP-MO) was performed on monkeys (Figure S2B). If the effect of LP-MO on CSF AD biomarker levels differed from that of the normal LP, it would indicate that CSF outflow had an effect on CSF AD biomarker levels. Conversely, if there was no difference, it would suggest that CSF outflow did not affect CSF AD biomarker levels.

Effect of LP-related deep intrathecal invasion on changes in AD biomarker levels

Deep intrathecal invasion of the spinal needle during LP may cause nerve irritation, which may stimulate the release of large amounts of Tau and NFL from the Tau- and NFL-rich cauda equina located in the lumbar region (Figure S2A).^26,27,28 Thus, to investigate whether changes in AD biomarkers were caused by deep intrathecal invasion, LP was performed under the guidance of real-time ultrasound imaging, allowing real-time observation of the needle location and minimizes intrathecal invasion (LP-MI, LP with minimal intrathecal invasion) (Figure S2C). If the effect of LP-MI on CSF AD biomarker levels differed from that of the normal LP, it would suggest that deep intrathecal invasion had an effect on CSF AD biomarker levels. Conversely, if there was no difference, it would suggest that deep intrathecal invasion did not affect CSF AD biomarker levels.

Effect of anesthesia before LP on changes in AD biomarker levels

As anesthesia is reported to potentially affect AD biomarker levels,^29,30 we also sought to determine whether this was a confounding factor in our study. Ruling out the effects of anesthesia in advance was not feasible due to the necessity of administering anesthesia prior to LP procedures. Therefore, we conducted an anesthesia-only procedure between the two LPs to investigate potential effects on CSF AD biomarkers. By comparing CSF AD biomarker levels before anesthesia-only (LP1) to those after anesthesia-only (LP2), any changes in levels post-anesthesia would suggest that anesthesia had an effect, while no changes would indicate that anesthesia had no effect.

We implemented rigorous experimental protocols to explore the effects of CSF outflow, deep intrathecal invasion, and anesthesia on CSF AD biomarkers. To ensure sensitivity in detecting changes in CSF AD biomarkers, we established a time frame for CSF sampling that corresponded to the peak effects of LP-MO, LP-MI, and anesthesia. As demonstrated in time-course study, the effects of LP on CSF Aβ and Tau levels peaked on day 3 post-LP, while the effects on NFL peaked on day 10 (see results). Thus, we conducted LP2 sampling 3 days after LP-MO, LP-MI, and anesthesia to investigate the effects on CSF Aβ and Tau and 10 days after LP-MO, LP-MI, and anesthesia to investigate the effects on NFL. We randomly assigned 45 monkeys into six groups, three to study the effects on CSF Aβ and Tau (Set A) and three to study the effects on CSF NFL (Set B). Detailed experimental protocols are as follows.

Set A. Experimental protocols for studying effects of CSF outflow, intrathecal invasion, and anesthesia on CSF Aβ and Tau levels

Effects of CSF outflow on CSF Aβ and Tau levels

As shown in Figure S3A, baseline CSF samples were initially collected from the LP-MO group monkeys (n = 7) in Set A (LP1). Subsequently, LP-MO was performed on each monkey after 10 days to effectively exclude the CSF outflow factor. Three days after LP-MO, when the effects of LP on CSF Aβ and Tau levels were maximal, LP2 was performed to obtain post-operation CSF samples. By comparing the CSF LP1 and LP2 samples, peak percentage changes in CSF Aβ and Tau levels after LP-MO relative to baseline were obtained. A different peak percentage change after LP-MO (LP2) from that before (3-day group) suggested that CSF outflow had an impact on CSF Aβ and Tau levels and vice versa (Figure S3A).

Effects of deep intrathecal invasion on CSF Aβ and Tau levels

As shown in Figure S3B, baseline CSF samples were initially collected from the LP-MI group monkeys (n = 8) in Set A (LP1). Subsequently, LP-MI was performed on each monkey after 10 days to minimize the intrathecal invasion factor. Three days after LP-MI, LP2 was performed to collect post-operation CSF samples. By comparing the CSF LP1 and LP2 samples, peak percentage changes in CSF Aβ and Tau levels after LP-MI relative to baseline were obtained. A different peak percentage change after LP-MI (LP2) from that before (3-day group) suggested that deep intrathecal invasion had an effect on CSF Aβ and Tau levels and vice versa (Figure S3B).

Effects of anesthesia on CSF Aβ and Tau levels

As shown in Figure S4, baseline CSF samples were initially collected from monkeys in the anesthesia group (n = 8) in Set A (LP1). Subsequently, anesthesia was performed on each monkey 10 days later to apply the anesthesia factor. Three days after anesthesia, LP2 was performed to collect post-operation CSF samples. The potential effects of anesthesia were determined based on the presence or absence of changes in CSF Aβ and Tau levels between LP1 and LP2 (Figure S4).

Set B. Experimental protocols for studying the effects of CSF outflow, deep intrathecal invasion, and anesthesia on CSF NFL levels

Effects of CSF outflow on CSF NFL levels

As shown in Figure 6A, LP-MO was performed on each monkey in the LP-MO group (n = 7) in Set B to exclude the CSF outflow factor. Since only a minimal amount of CSF (0.02 ml) was required for the NFL assay, we minimized the number of LPs to reduce animal suffering. The small amount of CSF sample (< 0.1 ml, equivalent to 6.7% of the 1.5 ml collected by normal LP) that inevitably leaked during LP-MO was collected as the baseline sample, instead of performing LP1 as in Set A. Ten days after LP-MO, when the effects of LP on CSF NFL were maximal, LP2 was performed on each monkey to collect post-operation CSF samples. By comparing the CSF samples collected during LP-MO (baseline) and LP2, peak percentage changes in CSF NFL levels caused by LP-MO relative to baseline were obtained. A different peak percentage change after LP-MO compared to that after normal LP1 in time-course study (obtained from 10-day group) suggested that CSF outflow had an effect on CSF NFL level and vice versa (Figure 6A).

Effects of deep intrathecal invasion on CSF NFL level

As shown in Figure 6B, LP-MI was performed on each monkey in the LP-MI group (n = 7) in Set B to minimize the intrathecal invasion factor. As mentioned above, to minimize suffering, CSF was collected during LP-MI to use as the baseline sample. Ten days after LP-MI, LP2 was performed on each monkey to collect post-operation CSF samples. By comparing the CSF samples collected during LP-MI (baseline) and LP2, peak percentage changes in CSF NFL levels caused by LP-MI relative to baseline were obtained. A different peak percentage change after LP-MI compared to that after normal LP1 in time-course study (obtained from 10-day group) indicated that deep intrathecal invasion had an effect on CSF NFL level and vice versa (Figure 6B).

Effect of anesthesia on CSF NFL level

As shown in Figure 6C, baseline CSF samples were initially collected from each monkey in the anesthesia group (n = 8) in Set B (LP1). Subsequently, anesthesia was performed on each monkey more than 100 days later (time-course study showed that the effects of an LP on NFL lasted less than 100 days). Ten days after anesthesia, LP2 was performed to collect post-operation CSF samples. The potential effects of anesthesia were determined based on the presence or absence of changes in CSF NFL levels between LP1 and LP2 (Figure 6C).

Due to the longer time required for the LP-MO and LP-MI operations and following veterinary recommendations, monkeys were deeper anesthetized to limit their movement with ketamine (7.5–15.0 mg/kg, intramuscular) combined with sodium pentobarbital (20.0 mg/kg, intramuscular), different from the normal LP procedure in which monkeys were anesthetized with ketamine only. The effects of this combined anesthetic on AD biomarkers were studied in two additional monkey groups. Results showed that ketamine combined with sodium pentobarbital had no significant effects on AD biomarkers (Figure S8).

To ensure an adequate number of subjects for this study, some monkeys were included in multiple experiments. However, only those monkeys that had not undergone any previous operations, including LP, for a significant period of time were selected to ensure that any previous effects were no longer present (See Table S1 for details).

LP sampling procedure

The monkey was fasted in the morning on the day of the LP procedure. After being anesthetized with ketamine (7.5–15.0 mg/kg, intramuscular), the lumbar region of each monkey was shaved, and the skin was disinfected twice with 3% iodophor solution. With the help of an assistant, the monkeys were placed in a seated position on the operating table and flexed forward to protrude the lumbar vertebrae slightly. An autoclaved 22 G Quincke spinal needle (Qionghua, Cat. No. GCZ-10, China) was inserted into the intrathecal space between L5/6 (L4/5 or L6/7 in 4.7% of cases). Once the stylet was slowly removed, 1.5 ml of CSF was collected in a sterile polypropylene tube (samples were not collected if contaminated with blood). After collection, the stylet was inserted into the needle, and the needle was removed, with a sterile cotton swab then pressed to the puncture site to prevent bleeding. To limit the influence of the 24-h biological rhythm, CSF samples from each animal used for before and after comparisons were collected at the same time of day. All procedures were performed under sterile conditions.

LP with minimal intrathecal invasion or CSF-outflow

To investigate the potential impact from CSF outflow and deep intrathecal invasion factors, we performed LP procedures under real-time guidance using an ultrasonic imaging system (ESAOTA, MYLAB 70VET XV, Italy) to track the position of the needle during LP. Each monkey was first anesthetized with ketamine (7.5–15.0 mg/kg, intramuscular) and then with sodium pentobarbital (Merck, 20.0 mg/kg, intramuscular) to ensure high-quality ultrasound imaging and accurate LP. The monkey was positioned on the operating table in a prone position with hind legs curled under the abdomen. The lumbar area was shaved and prepared with 3% iodophor solution. A sterile puncture rack (Jingfang, JSM-113, China) was installed on the probe to ensure that the mounted sterile spinal needle moved in the plane of the ultrasonic beam. A paramedian oblique sagittal scan of the lumbar spine was performed to locate the L5/6 lumbar interspace. The needle was inserted in the plane of the ultrasound beam from the caudal end of the probe with the tip directed towards the dura mater of the L5/6 lumbar interspace, allowing the advancing needle to be tracked in real time. Once the needle reached the dura mater, two types of operation were performed according to the experimental protocols.

LP with minimal intrathecal invasion (LP-MI)

Deep intrathecal invasion of the spinal needle during LP may cause nerve irritation, potentially may stimulating the release of large amounts of Tau and NFL from the Tau- and NFL-rich cauda equina located in the lumbar region. We speculated that depth of intrathecal invasion was a key factor in the elevation of AD biomarkers. Thus, to minimize intrathecal invasion, once the needle passed approximately 1 mm through the dura mater, further insertion was stopped to avoid deeper penetration and potential injury to the cauda equina nerve. After withdrawing the stylet, 1.5 ml of CSF was collected in a sterile polypropylene tube, followed by reinsertion of the stylet into the needle. Subsequently, the needle was removed, and sterile gauze was applied to the puncture site to prevent bleeding. The whole procedure was performed under sterile conditions.

LP with minimal CSF-outflow (LP-MO)

In this operation, after the spinal needle has passed through the dura mater, it continued to be inserted to the center of the intrathecal space. The stylet was withdrawn to let a small amount of cerebrospinal fluid (CSF) flowed out to confirm that it had reached the intrathecal space. To minimize the CSF outflow, a CSF volume of less than 0.1 ml (equivalent to 6.7% of the 1.5 ml collected by normal LP) was collected in a sterile polypropylene tube, with the stylet immediately inserted back into the needle to terminate CSF outflow. Subsequently, the needle was removed, and sterile gauze was applied to the puncture site to prevent bleeding. The whole procedure was performed under sterile conditions. In addition, the LP procedure causes dura mater rupture in humans, potentially leading to CSF leakage into the epidural space through the rupture site which will cause CSF outflow. If monkeys experienced a similar CSF leakage after LP, then the LP-MO operation in mechanistic study could not minimize the CSF outflow factor effectively. Thus, to examine potential CSF leakage after LP-MO in monkeys, we performed magnetic resonance imaging (MRI) scans of the lumbar region at 1 day after LP-MO. No CSF leakage was detected, as determined by a professional radiologist. Thus, the LP-MO operation successfully minimize the CSF outflow factor (Figure S9).

CSF sample processing

All collected CSF samples were placed on ice and immediately centrifuged (1 800 g, 10 min, 4°C). After centrifugation, the samples were aliquoted into 0.5-ml sterile polypropylene cryogenic vials, quick frozen in liquid nitrogen within 30 min, and transferred to a refrigerator at −80°C for storage until analysis.

CSF biomarker analysis

The CSF samples were thawed at 4°C just prior to analysis. Monkey CSF t-Tau (Figure 2), Aβ40, and Aβ42 levels were measured using Human Tau (Total) (Thermo Fisher, Cat. No. KHB0041), Amyloid-beta^{1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40} (Thermo Fisher, Cat. No. KHB3481) and Amyloid-beta^{1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42} (Thermo Fisher, Cat. No. KHB3441) ELISA kits according to the manufacturer’s instructions. Monkey CSF t-Tau (Figure 5), p-Tau181, p-Tau231, and NFL levels were quantified using ultra-sensitive Simoa technology (Quanterix, MA, USA) on the automated Simoa HD-X platform (GBIO, Hangzhou, China) according to manufacturer’s instructions. Human Tau (Cat. No. 101552), P-Tau 181 V2 (Cat. No. 103714), P-Tau 231 assay (Cat. No. 102292), and NFL (Cat No: 103186) kits were purchased from Quanterix and used accordingly. Samples collected from the same monkey for before and after comparison were run together on the same 96-well plate and tested with the same batch of reagents.

Quantification and statistical analysis

Data analysis was conducted using GraphPad Prism v8.0.2. Grubbs’ Test (Alpha = 0.01) was applied to identify outliers in the change rate (% change = [LP2 − LP1] / LP1 × 100%) of LP2 relative to LP1 for each group. Outliers as well as the LP1 and LP2 data corresponding to the outliers were discarded (except for p-Tau181 in the 5-day group, which changed to “ns”, no changes in the significance in other groups after the outliers were deleted were detected). Two-tailed paired t-test was used to compare LP1 and LP2 means of each group. Significance was set to p < 0.05. If the data failed the Shapiro-Wilk test for normality, a Wilcoxon signed-rank test for paired data was conducted. To compare the percentage change in Part II, unpaired t-test for independent samples was conducted. The Mann-Whitney U test was used to compare percentage change when the data failed the Shapiro-Wilk test for normality, or when the F test (p < 0.05) showed uneven variance. ns: p > 0.05, ∗: p < 0.05, ∗∗: p < 0.01, ∗∗∗: p < 0.001, ∗∗∗∗: p < 0.0001.

Published: March 6, 2024