We thank Dr. Bernards for the interest in our study. He questioned the validity of the model used in our recent study examining cerebrospinal fluid leak rates and dural trauma patterns produced by epidural needle puncture of human cadaveric dura1and of models used in similar work with spinal needles. He states that these models are inadequate to study the “mechanism for spinal headache or the methods that can be used to prevent it” because such models produce punctures of the dura and not the spinal arachnoid membrane. He further states that in vitro models cannot account for the healing process or identify the “rate limiting meninx,” and that extrapolation from these bench models cannot be made to the clinical setting.
He is correct in stating that we assume that cerebrospinal fluid leak is associated with postdural puncture headache. There is considerable consistent evidence from diverse sources in the broader medical literature to support a “causative” relationship between cerebrospinal fluid loss and headache as well as an association between greater loss and an increasing incidence and severity of postdural puncture headache.2–8
We do not claim to have examined the mechanisms of headache itself but rather mechanisms influencing its potent trigger, cerebrospinal fluid loss. The mechanisms responsible for postdural puncture headache symptomatology are likely to be far more complex than traditional teaching would suggest and not amenable to study using an in vitro model.
The existing literature supports the assumption that spinal needle gauge and tip design play a significant role in postdural puncture headache, with smaller gauge and pencil tip needles leading to a lower incidence of headache.5–6,In vitro work also suggests a correlation between leak reduction and spinal needle design.7–8It is logical to assume that epidural needle gauge and tip design may also impact on postdural puncture headache after unintentional puncture. The absence of clinical trials comparing the effect of epidural needle design on postdural puncture headache prompted our in vitro study.
The spinal arachnoid membrane consists of a laminar portion that forms a watertight lining attached to the dural undersurface and a trabecular portion that extends from the laminar arachnoid to the pia mater.9–10Cerebrospinal fluid flows within the space (subarachnoid space) contained between the laminar arachnoid and the pia. Dural puncture involves passage of the needle through the external dura, the capillary interval/potential space known as the subdural space, and the laminar arachnoid membrane, which contains cerebrospinal fluid. The subdural space is not usually a prominent feature in these tissue layers and appears to be opened by trauma, fluid injection, and possibly by capillary proliferation (versus engorgement) in reaction to dural injury involving cerebrospinal fluid loss.9–11
In our study, fresh human cadaveric lumbar spinal cords were removed at autopsy with meninges intact and cooled in lactated Ringer’s solution. Full thickness dural specimens (including attached laminar arachnoid membranes) were then harvested for study. The laminar arachnoid was easily visualized at the time of dissection as a smooth shiny membrane closely applied to the dural undersurface in all of the cadavers studied. Care was taken not to traumatize the dura arachnoid interface during tissue handling or mounting on the model.
We examined leakage of artificial cerebrospinal fluid and dural tissue trauma patterns by scanning electron microscopy after standardized puncture of specimens with epidural needles. The model used in our work is similar in many respects to models used in spinal needle studies. Tissue was mounted over a window in the model and sealed with a customized gasket containing a matching window through which punctures were performed. Features of our model that add to its clinical relevance include 1) the use of a physiologically pressurized cylindrical model with a diameter closely approximating that of the lumbar dural sac; 2) mounting of dural specimens with in vivo orientation maintained; 3) use of artificial cerebrospinal fluid; and; 4) observed tenting of the dura at the time of puncture.
Our model was watertight (as were similar spinal needle models) when pressurized to physiologic pressures with artificial cerebrospinal fluid and, save for isolated leakage from deliberately made needle puncture sites, remained that way. Demonstration of a watertight seal with dural specimens in such models suggests that the laminar arachnoid membrane (with tight junctions)9was not only present but also intact, providing a barrier to fluid passage. It should be noted that authors using similar watertight models for spinal needle studies have specifically noted the presence of the arachnoid membrane on their dural tissue specimens, with more recent studies specifically demonstrating the presence of laminar arachnoid membrane on the inner surface of dural punctures using scanning electron microscopy.7,12
The assumption that the pattern of needle injury to the external dural surface bears no resemblance to underlying arachnoid injury is inconsistent with our observations in the laboratory and it is not supported by the limited existing literature.12As the laminar arachnoid covers the dural undersurface as a thin membrane, one would expect the morphology of needle injury to the external dura to closely approximate that on its inner laminar arachnoid surface. We observed that this was the case on gross examination of the internal and external surfaces of our dural specimens. Reina et al. 12also demonstrated this finding in scanning electron microscopy images taken of both the laminar arachnoid (internal) and external surfaces of dural specimens after single spinal needle punctures. The authors showed that the pattern of needle injury was similar regardless of the dural surface imaged. They also found that the calculated area of a single puncture site was not significantly different regardless of whether it was imaged and measured on the laminar arachnoid or external dural surface.
Given this information, we chose to examine injury patterns with scanning electron microscopy from the external dural surface of our specimens because cerebrospinal fluid leaks not only through the hole in the laminar arachnoid but also through the channel produced in the wall of the fibrous dural sac itself. Our findings suggest that one should not discount the fibrous dural sac, dural tissue fragments, and hole morphology as potential modifiers of leak rates after puncture. The presence of several specimens in our experiment that failed to leak or leaked slowly after obvious epidural needle puncture and subsequent demonstration of full or partial occlusion of these sites with dural tissue fragments (via scanning electron microscopy) may help to explain the absence of postdural puncture headache in some patients following obvious unintentional dural puncture. Plugging also presents a likely explanation for observations in spinal needle studies through the years that puncture sites may cease to leak or not leak at all following spinal needle withdrawal.
Traditional teaching that the dural fibers are oriented in a parallel fashion has not borne the weight of scrutiny when subjected to more powerful imaging techniques such as scanning electron microscopy13and transmission electron microscopy (our own unpublished data). Given inconsistencies in the results of in vitro studies examining leak rates after parallel versus perpendicular bevel orientation during puncture14–15and the small number and limited methodological quality of clinical studies examining postdural puncture headache in this setting,16–18we believe that it would be more reasonable to say that there is little high quality evidence to support a difference in the incidence of postdural puncture headache as a result of needle bevel orientation at the time of puncture.
Our model does not address the process of dural healing. It is implied in Dr. Bernards’ letter that the absence or early resolution of postdural puncture headache in some patients is secondary to early dural or arachnoid healing. We would consider this to be unclear at best. We would suggest that the weight of the existing, albeit lower quality, evidence in the literature would support that this is not the case, especially for larger gauge punctures.
Few studies have addressed the issue of dural healing. Franksson and Gordh,19examined the postmortem dura of three patients with documented lumbar punctures at 2 days, 14 days, and 40 days before death. At 2 days, no evidence of healing was found, and at 14 days, only early signs of healing were evident at the corners of the dural tear made by a bevelled needle. At 40 days a scar was macroscopically visible in the puncture site, with microscopic evidence that the tissue filling the site was of recent origin. Leaking dural defects have also been noted during surgery after “protracted periods of time” have passed following lumbar puncture.20–21These findings along with a relatively low permanent cure rate of initial epidural blood patches,22–23and case reports of successful but late (weeks to months) or late repeated use of blood patching for postdural puncture headache are also consistent with a relatively slow process of dural repair.
Given evidence suggesting a slow healing process, that the incidence and severity of postdural puncture headache is likely to be related to the volume and rate of cerebrospinal fluid loss, and that this in turn is very likely to be related to tissue injury with needles, we maintain that our in vitro model and the models of our counterparts remain relevant to the study of postdural puncture headache.
* University of Toronto, Sunnybrook and Women’s College Health Sciences Center, Toronto, Ontario, Canada. firstname.lastname@example.org