Meanwhile, the vascular system is developing the four-columned foundation of a carotid artery on each side of it and a pair of vertebral arteries that feed the brain. At the top of the brain stem, the four arteries meet and feed a circular arterial system that then branches off to supply different parts of the cortex, brain stem, and cerebellum.
The muscular walls of the arteries will control blood flow and pressure in the brain by contracting and narrowing the blood vessels, like a hand pinching off a garden hose; blood flow and pressure beyond the contraction point diminishes. When fully developed, this system will allow blood to be preferentially sent to the brain in times of oxygen deprivation (the body prioritizes the most important organs—the brain, the heart, and the adrenal glands).
But early in development, this pressure-control system is immature, and blood flow in the brain resembles fluid going through a lead pipe rather than through a garden hose: pressure in equals pressure out.
In the fluid-filled lateral ventricles, one of which lies under each side of the cortex, the blood supply develops into a meshwork of tiny arteries called the germinal matrix. During the nine months of gestation, the germinal matrix evolves from a frondy network floating within the cerebrospinal fluid in the ventricle to a thin hardy strip of blood vessels that’s firmly anchored to the ventricle’s wall.
The combination in the immature brain of insufficient pressure control and weak germinal matrices floating inside the ventricles can be a lethal one if the baby is born too early. Blood pressure shifts wildly during a premature delivery, and the pressure within the brain can rise and fall sharply if uncontrolled by the arterial system, causing the fragile germinal matrix to burst and spew blood into the ventricular cavity. Named the germinal matrix because it gives rise to many of the hemisphere’s neurons, this critical system is weak in the premature period only, an example of how biologic structures that are resistant to minor insults in maturity can nonetheless be particularly vulnerable to injury during development.
Think of an intraventricular hemorrhage as an explosion—one whose strength and consequences won’t be known until later. The IVH can be mild—a little blood that leaks into the ventricular space that might not cause any long-term injury. These bleeds are given a 1 or a 2 on a scale of 1 through 4.
The blood from a more significant IVH fills the ventricle, clots, and then blocks the narrow passages that allow cerebrospinal fluid to circulate. With blood filling the ventricular space, and cerebrospinal fluid building up, pressure within one or both ventricles increases, which is when the real trouble begins. This is a grade 3 bleed.
The blood and pressure begin to strangulate the cells lining the ventricles—the cells that are supposed to develop into insulating white matter. The brain is compressing itself from the inside out. If bleeding from the tiny ruptured arteries continues, the pressure inside the ventricle builds, and the outward force in the ventricle can equal or exceed the arterial pressure supplying blood to the brain, cutting off the circulation and preventing surrounding brain tissue from receiving oxygen.
Once dead, these white-matter precursors don’t regenerate. Over time, the dead cells liquefy, leaving clear spaces like the holes in Swiss cheese. The result is a slowdown in conduction from the cortex to the affected body parts—often arms and legs—so movement is slow or nonexistent, and the weakened signals from the cortex must fight the natural tendency of muscles to contract. This type of injury to the connection from brain to body is known as cerebral palsy.
Larissa’s first ultrasound showed blood in both ventricles, more in the left than the right. But it also showed bleeding outside the ventricle on the left side—within the brain tissue itself.
This is referred to as a grade 4 bleed—the worst kind—because the bleeding doesn’t just threaten injury to surrounding brain cells, it causes direct and permanent injury to brain cells in the vicinity of the bleeding.
The 1 through 4 grading system works well from the perspective of the consequences of the bleeding—grade 1 bleeds rarely have long-term consequences, and grade 4 bleeds always do.
Bleeds on the continuum from grade 1 to grade 3 represent blood in the ventricles. Grade 1 describes a bit of blood in the ventricle; grade 3 describes blood that clogs and causes dangerous swelling within the ventricular system. Some neurologists, however, believe that the grade 4 bleed results from a different mechanism because it is typically caused by bleeding from vessels other than the germinal matrix and because injury results from direct cell destruction, not from the pressure and irritation of the ventricle pushing outward. Ringer felt that, in Larissa’s case, the injury was caused by trauma during her delivery.
Although the outcomes of babies with these severe hemorrhages have become somewhat better in recent years, when Larissa was born, the report of the definitive study of seventy-five babies with grade 4 bleeds like Larissa’s was dismal: 59 percent of them died, and of the twenty-two children who survived to be evaluated later, 87 percent had major movement disabilities, and 68 percent had significant cognitive impairment.1
But brain injury in these babies isn’t just a matter of grading. Why do some babies with a grade 3 IVH develop severe cerebral palsy while others with the same type of bleed exhibit only minor clumsiness? Part of the answer may be luck, but other factors clearly influence outcome. If a baby’s blood pressure bounces up and down during the first hours or days of life, the brain may spend periods of time without enough oxygen, causing long-term damage to the brain cells. Infection can also influence the injury. The body’s response to infection can lead to the dumping of chemicals called cytokines, which actually injure nearby cells—and if this occurs in the same area where bleeding has caused injury, the effect can snowball.
The unknown in Larissa’s injury was the suggestion that bleeding had occurred in her cerebellum. When Larissa was born, there were few publications that described the long-term outcomes of cerebellar hemorrhage, but the importance of that part of the brain for the coordination of movement was well understood, and there was an emerging understanding among neurologists that the cerebellum also had an enormous role in learning, behavior, and information processing. A few years after she was born, a report on more than fifty children who had had cerebellar hemorrhages showed that half of them had movement impairment and a third had abnormal cognition, communication and behavioral abnormalities, and even autism.2
An hour after Larissa returned to the Brigham NICU, I put my white coat on over a pair of scrubs and walked across the bridge to Children’s Hospital. I hadn’t been in the hospital since I was a child, and coming in the back entrance, I found the place unfamiliar and disorienting.
“Hi, can you direct me to the MRI reading room?” I asked a secretary. My heart was racing, but I tried to affect the bored look of a tired resident.
“Through that door, right at the bottom of the steps,” the secretary answered without really seeing me.
I followed her directions and pushed into a small, crowded darkened room stacked with video monitors. The smell of old coffee, potato chips, and poor ventilation wafted up. No one looked at me.
As my eyes adjusted to the light, one of the residents reading imaging studies called out, “Can I help you?” Neither he nor the three or four other residents and medical students who surrounded him glanced up.
“Yeah,” I said nonchalantly, “I’m one of the OB residents from the Brigham, and one of our babies just had an MRI. Would you mind giving me a wet read?”
“Sure, no problem. Name?”
“Lowery. Baby Lowery.”
“What’s the story?”
“Twenty-six weeks. Now day of life four. Ultrasounds over the weekend showed bilateral bleeds and maybe a cerebellar hemorrhage as well.”
There was a pause as the resident brought the study up on his dual-monitor display and scrolled through the hundreds of two-dimensional images of Larissa’s brain slic
ed in all three planes by the MRI.
“Okay. Let’s see. There is a left ventricular bleed, and also a moderate-size left parenchymal bleed,” he said, describing the bleeding in the brain tissue adjacent to the left ventricle. “There is a little bit of blood in the right ventricle, but no evidence of bleeding. I think it may have come across from the left side. Both ventricles are normal size.”
He turned toward the group sitting around him and peering over his shoulder. “You know, that’s interesting. Here we have a hemorrhage that looks three to four days old and involves much of the left ventricle and the parenchyma around it, and yet the ventricle isn’t dilated. That is unusual.”
“What about the cerebellum?” I asked, trying to sound like this was almost an afterthought.
“Oh, yeah. Cerebellum looks pretty good. I see a small area here,” he said, pointing. “Maybe there is a small bleed here. I’m not a hundred percent sure. I’ll have to ask my attending, but if there is a bleed, it’s a small one.”
For the first time he turned toward me. “Anything else?”
“That’s it,” I said. “Thanks.”
“Good luck,” the resident said before clicking on the name of a new patient to bring up the images on his screens.
I walked back toward the Brigham to tell Kelly what I had heard.
Later that day, we entered the NICU to find Larissa completely surrounded by men and women in long white coats. The neurologists had arrived. Their attire was formal and their manner was somber. They were in the process of examining Larissa.
They measured her head circumference; they shone a flashlight into her eyes; they tested her primitive reflexes, pulling against her tiny arms and letting them spring back toward her to see if she retracted them equally, and then doing the same thing with her little legs.
They methodically flipped through her chart and wrote down her laboratory values. Then, in what seemed like single file, they walked to a computer monitor that allowed them to review the ultrasound images and MRI scans taken earlier that day. They spent some time arguing over whether there was a cerebellar hemorrhage or not, and they ended up agreeing that there probably was a small one, like the radiologist had said. Then the lead neurologist wrote a single sentence in the chart: Neurology consult to follow. And the line of doctors trooped out of the NICU and back to Children’s Hospital so they could spend some time pondering what the future would hold for my daughter.
Already, four days after her birth, Larissa’s brain injury was reasonably well defined, although it wasn’t entirely clear if she had a cerebellar injury. If Larissa were an adult who’d had a similar injury as a result of a stroke or a traumatic event, the outcome would be predictable. You may remember the homunculus from high school science—the map of the body drawn on the brain. Scientists have mapped the regions of the brain that typically control specific parts of the body. Using this map and comparing it to the site of injury based on the ultrasound and MRI findings, neurologists could predict the extent of an injury.
But Larissa wasn’t an adult—she wasn’t even a full-term baby. Although Larissa had completed two-thirds of the length of a normal gestation, brain development occurs toward the end of pregnancy. At twenty-six weeks, the brain weighs only about 30 percent of what it will weigh at term, and because it is largely smooth, in contrast to the complex gyrations it will achieve later on in development, it has only about 20 percent of the surface area of what it will have at term. Whether Larissa’s remaining brain growth would continue on a relatively normal trajectory—and the brain develops better in utero than it does in the NICU even under the best of circumstances—or be adversely influenced by the hemorrhage was entirely unknown.
The most important unknown—the question whose answer would determine Larissa’s future—was the extent to which the unscathed brain tissue on the right would take up the tasks meant for the brain tissue on the left that had disappeared into an ugly glob of hemorrhage. Larissa’s future would be determined by neuroplasticity.
The brain works by building and then maintaining or neglecting connections among neurons. For example, during development, a neuron that originates in the area near one of the ventricles can attach itself to cells near the brain surface and, using surrounding white-matter cells for guidance, build a scaffolding within the cell to support outward growth. Cells near the brain surface secrete chemicals called neurotrophic factors that serve as homing beacons for cell growth.
Neurons communicate with one another using action potentials: electrical charge builds up inside the cell and then, when a critical threshold is reached, travels down the length of the neuron’s axon. The axon is the expressive part of the neuron, and its end lies in very close proximity to the receptive part of another neuron, called the dendrite; the axon and dendrite are separated by a very small space, called the synapse.
When an action potential travels down an axon and reaches the end, the axon releases a type of chemical called a neurotransmitter into the synapse. The neurotransmitter fits, like a key into a lock, in receptors on the dendrite of the neighboring neuron. The combination of neurotransmitter and receptor opens gates in the cell wall, and charged molecules flood in and create an action potential in the neighboring cell.
I am simplifying this next part of the process dramatically, but imagine that an action potential has fired in a neuron in the part of my brain that handles expressive language. From one neuron to another, action potential to neurotransmitter to action potential, the signal is relayed to my left motor cortex, the region of the cortex that handles voluntary movements. Here, the signal is passed to a neuron that extends all the way down the spinal cord, crossing from the left to the right side at the base of my skull. Then the signal is transmitted to yet another neuron, one that stretches the entire length of my right arm, all the way to a specific muscle in my right ring finger. At that muscle fiber, the axon releases acetylcholine, a neurotransmitter that doesn’t cross to another neuron’s dendrite but instead triggers the firing of the muscle that makes my finger contract and hit the period key on my keyboard that’s needed to end this sentence.
When I was learning to type, the process was clumsy and slow. Over time I got faster and increasingly accurate. In my brain, there was no proliferation of new neurons, but rather an increase in the density of the receptors on the neurons involved in this circuit. With each new receptor added, the amount of neurotransmitter needed to trigger the action potential diminished. This is how learning occurs.
The homunculus map that links a specific part of the body and its function to a single, isolated region of the brain originally came from research that indicated that the map was an immutable system—that is, once a function was localized to a particular part of the brain, the location of that function would never change. Every medical student learns about Broca’s area, a section of the left frontal lobe associated with the motor control of speech. The function of the area was described by the physician Paul Broca in 1861 after a patient who had lost the ability to speak was found on autopsy to have a syphilitic lesion in the left frontal lobe. In Broca’s patient, the injured frontal lobe couldn’t accomplish the “thought” that initiated speech—the connection between the brain and the muscles of the mouth and tongue worked just fine, but no signal initiated speech. The fixed nature of the homunculus—locationism theory—became dogma, and for nearly a century, any suggestion that the fixed system wasn’t all that fixed was ignored.
In the 1960s and 1970s, research started to chip away at this orthodoxy. Experiments on monkeys showed that when a brain was rewired—when the optic nerve was surgically directed toward the auditory cortex instead of the visual cortex, for example, or when nerves from different fingers were surgically swapped on their way to the brain—the change was accommodated. The auditory cortex, typically the area that processes sound, learned to process visual input. The brain relearned the location of the fingers
to allow for normal hand function.
Clinical experience was undermining locationism theory. In his fascinating 2007 book on neuroplasticity The Brain That Changes Itself, the psychiatrist Norman Doidge described the prolonged and intensive at-home rehabilitation of Pedro Bach-y-Rita, a man who had a devastating stroke that left him paralyzed. Instead of allowing his father to waste away in bed, Bach-y-Rita’s son worked with him for hours each day. He re-taught his father to crawl, kneel, and finally walk, and, by repeatedly practicing crude motions with his hands, arms, and fingers, his father slowly regained normal function. Pedro Bach-y-Rita’s son, Paul, happened to be a leading scientist, and he was as responsible as anyone for reviving the science of neuroplasticity in the 1970s. When Pedro Bach-y-Rita finally died, of a heart attack while mountain climbing in Colombia, his autopsy showed that 97 percent of the connections between his cortex and his spinal cord had been destroyed—he had regained complete function of his arms and legs by teaching the remaining few neurons to take over the tasks of their destroyed neighbors.
Back at Children’s Hospital, the neurologists reviewed Larissa’s MRI and then sat down in a conference room to discuss their findings. The medical literature predicted awful outcomes for children with grade 4 intraventricular hemorrhages like the one Larissa had. But there was anecdotal evidence suggesting that this wasn’t always the case.
Adre du Plessis, a brilliant and profane neurologist originally from South Africa, had a particular interest in premature-infant brain injury, and he’d been collecting cases for a report on grade 4 intraventricular hemorrhages. His research showed that kids with unilateral grade 4 hemorrhages often weren’t as devastated as those with the more common bilateral type of hemorrhage. He thought that the reports his colleagues were reading were crude, and the awful outcomes might be influenced by the larger number of bilateral hemorrhages in the studies. One of his research fellows had analyzed the group with unilateral bleeds and found that, as he put it, “The numbers were a hell of a lot better” than what had been described. “Functionally,” he said, “these kids were in the game.”
Fragile Beginnings Page 7