Path. Res. Pract. 191,924-934 (1995)
A Unifying Hypothesis that Links Benign Prostatic Hyperplasia and Prostatic Intraepithelial Neoplasia with Prostate Cancer Invited Comments H. Harvey Puckaster Close, Puckaster Lane, Niton Undercliff, Ventnor, Isle of Wight, England
SUMMARY Prostatic Intraepithelial Neoplasia (PIN) and prostatic cancer (PCA) are not caused by infection, allergic reaction, inadequate immunological response, ischemia, ageing, systemic hormones, carcinogens, nor prostatic ductal contents. PIN and PCA are apparently caused by increased inner acinar pressure due to partially blocked draining ducts. Only this explanation can account for all the observations about PIN and PCA. All other possible causes are disproved by specific observations. In order to further clarify the cause of PIN and PCA, it is important to discover if peripheral zone lesions cluster around ducts or blood vessels. PIN patterns are the morphological precursors of both PCA and prostatic cysts. Different PIN patterns represent different adaptive stages to increasing inner acinar pressure. The immediate tissue cause of PCA is PIN disruption seeding the stroma with high-grade PIN (HGPIN) cells. These cells, programmed for adaptive proliferation and mobility in PIN, are sufficient in the stroma to cause all stages and patterns of invasive PCA. No mutated cells are necessary. For reasons given, the primary cause of the initial ductal blockage that results in PIN and PCA cannot be inflammation, stones, proteineous plugs, infarction, venus thrombosis, ductal hyperplasia, nor a constricted penis at ejaculation. Only benign prostatic hyperplasia (BPH) can explain all the facts and is thus the primary cause of the ductal blockage resulting in cysts, PIN and PCA. The main causes of BPH are apparently disuse atrophy of sexual and abdominal muscles, and atherosclerosis of the capsular branch of the prostatic artery, causes atypical adenomatous hyperplasia (AAH) in the transition zone. The resulting muscular and glandular atrophy decreases local and general growth inhibitors. New growth in the adult prostate is abnormal because epithelial cells grow into ducts rather into the stroma. In such ducts, the growths cannot receive stromal growth inhibitory signals, and thus continue to grow indefinitely and result in BPH, AAH-adenosis, blockage of ducts, cysts, PIN and PCA.
© 1995 by Gustav Fischer Verlag, Stuttgart
International Consultation on PIN - Invited Comments . 925
Montironi et al., Silvestri et al., Polito et al., Arakowa et al., Van de Voorde and Qian et al.. Neither PIN nor PCA can be primarily caused by infection nor allergic autoimmune reaction because none of the early or late lesions are consistently associated with inflammatory cells such as neutrophils, lymphocytes, or macrophages. Nor do we find any of the usual systemic reactions that accompany chronic infections or allergy. Also, if infection or an allergic reaction was a major cause of PIN and PCA, we would expect to find more lesions closer to the urethra (the likely portal of entry of noxious agents) in the transition zone and in the periurethral zone, rather than so many lesions deep within the glandular tissue of the peripheral and central zones. Also, we would expect to find more PIN in twenty to thirty year old men, since these are the years of the greatest exposure to the likely noxious agents. On the contrary, we find the maximum lesions in aged men. Thus, neither infection nor allergic reactions are the primary cause of PIN and PCA. Ischemia, though common in the peripheral zone (PZ) of the aged prostate, is rarely spatially close to PIN and PCA, which, on the contrary, are exuberantly growing lesions with a prominent blood supply. Also, these lesions do not have the cellular reactions expected of ischemia, nor do they have the wedge shaped tissue found in ischemic infarction. In fact, Montironi et al. report that high grade PIN (HGPIN) and PCA have a greater than normal density of capillaries. Thus, ischemia is unlikely to be the primary cause of PIN or PCA. Ageing is not a major cause of PIN or PCA because, as Silvestri et al. and Sakr et al. report, both lesions are often found in relatively young thirty to forty year old men, and many elderly men have neither lesion. This is unlike ageing when universally associated with skin wrinkling. Though PIN and PCA increase with ageing, so do nearly all adult cancers. So ageing is not directly itself causal of these lesions. An inadequate immunological response is not a significant ca use of PIN or PCA because both are found in men who have no other signs of immunological compromise, and neither prostatic lesion appears increased in patients who are immunologically compromised (for example, in those with AIDS or in the immuno-suppression of transplantion). Are increased systemic hormones a primary cause of PIN and PCA? None of the usual hormones (e.g. testosterone, estrogen, prolactin or progesterone) have been consistently associated with PIN and PCA in human studies. As reported by Pylkkanen et al., all studies that have treated laboratory animals with hormones (usually testosterone and/or estrogen) to produce PCA have had to use abnormally high doses of hormones, and the dose age schedules have always been non-cyclic. Men normally have daily cyclic variations in androgenic hormone blood levels due to nightly increase in melatonin. Such variation has never been controlled for in Unlikely Causes of PIN and PCA any laboratory animal resulting in PIN or PCA. Also, PIN has been shown to be a precursor of PCA, as re- the lesions in animals are almost always in the prostatic ported at this meeting by Bostwick et al., Rodolfo ducts and not in acini, as in man. Further, all hormones
Because PIN occurs so early in the genesis of PCA, understanding it is pivotal for diagnosis, treatment, prognosis, screening, prevention and causal analysis. I want to set each contribution of this conference in a dialectic of evidence and arguments for and against the possible causes and pathology of PIN and PCA . My approach to PIN and PCA has been mainly theoretical, and, similar to a purely theoretical astronomer, I must rely on all the significant research by bench and clinical scientists as the foundation of the causal explanations of PIN and PCA which I construct, evaluate, tear down and rebuild. The scientific methodology is Popperian and Bayesian, in that at each causal junction, all possible causes must be evaluated with emphasis on disproof of a possible cause as the best selection and elimination principle. For example, we can never prove that all swans are white, but we can easily disprove this by finding one black swan. Disproof of some possible explanations increases the probabilities of the remaining possible explanations, and the likely true explanations are those that withstand repeated attempts to disprove them by drawing out all their implications - especially those implications that are unique predictions that form the basis of crucial experiments. An example in astronomy is the unique relativistic prediction that the sun would bend nearby light rays. Since nearly any such disproof can have an ad hoc explanation to help revive a disputed causal hypothesis, all disproofs themselves must be repeatedly re-examined, tested and carried forward as part of the ultimate data and reasoning supporting any concluding causal explanation. The following analysis uses most of the reported findings of this conference to show the primary cause of PIN and PCA cannot be solely infection, allergic reaction, inadequate immunological response, ischemia, ageing or systemic hormones or carcinogens, prostaglandins, lipids, macrophages or neutrophils. Rather, a necessary and sufficient explanation of the cause of PIN and PCA is the increased inner acinar and ductal pressure due to partially blocked downstream ducts. No mutations are necessary. Such blocked ducts and the resulting acinar adaptive hyperplasic responses appear to explain all the relevant available data about PIN and PCA, and to make many unique predictions. This suggested explanation has implications about prevention, screening, diagnosis, treatment, and prognosis. The primary motivation in presenting the following analysis is to receive help in evaluating this reasoning. Has any relevant explanation been overlooked? Have any relevant data been omitted? What other consequences follow from the assumptions? Are the unique crucial predictions true?
926 . H. Harvey
produce prostatic lesions only in highly selected strains of animals, and most other strains or out-breed animals get few or no resulting lesions from the same treatment. Thus, I conclude that increased systemic hormones levels are not an initiating cause of PIN or PCA in man, and, at best, hormones are only permissive, as are necessary nutrients. But even this is questionable, since PIN and PCA increase with increasing age, whereas testosterone decreases with increasing age. The one study in man that found increased androgens in Afro-Americans, who also have increased PIN and PCA, as evidenced in the report here by Sakr et al., found this hormone only in young men at an age when increased aggressiveness could explain these findings. That study did not control for diet or other relevant variables that are associated with hormone levels. Systemic chemical carcinogens as initiating causes of PIN and PCA have many of the difficulties listed above for hormones: no consistent findings in human studies, animal studies with abnormally high doses, only found in carefully selected strains of animals, or the resulting animal prostatic lesions originate in ducts and not in acini. Additionally, cigarette smoking, which exposes man to high doses of nearly all the relevant chemical carcinogens that produce prostatic lesions in animals studies, has not been associated with increased PIN or PCA in about fifty retrospective and prospective human studies. The one study that found smoking associated with PCA, with a risk factor of about two, did not correct for crucial promotional risk factors such as diet or hormone levels. Thus, I conclude that systemic chemical carcinogens are not a significant initiating cause of PIN and PCA in man. Observations about PIN suggest another argument against systemic hormones and carcinogens as primary initiating causes, also against any possible systemic cause of these lesions. Peripheral zone (PZ) clusters of PIN usually appear to have a lobular orientation; the cluster is usually round or oval in outline, often with a large draining duct somewhere near the center of the cluster, and separate similar clusters are often found at distant parts of the PZ, even on the opposite side. Such observations suggest that clusters of PIN arise around and from ducts, and separate clusters arise from separate ducts. Such a distribution of PIN clusters is not what we would expect if PIN were initiated by a causal factor supplied by the blood vessels, which in the PZ originate peripherally from the capsular branch of the prostatic artery. If this blood supply delivered the primary initiating cause of PIN and PCA, as in ischemia and infarction, we would expect PIN acini to cluster, not around a duct from the central urethra, but around a vascular bundle from the prostatic capsule. Also, if blood supplied the initiating cause of PIN and PCA, we should be surprised to find, as we do, the greatest amount of PIN and PCA in the PZ of the aged prostate which also has the greatest amount of ischemia, infarction and atrophic acini and ducts due to atherosclerosis of the capsular branch of the prostatic artery. Further, if blood supplied the initiating cause of PIN, we would
expect a cluster of PIN acini not to respect a lobular orientation around a single individual duct. We would rather expect to find PIN in acini representing different adjacent draining ducts, since the vascular bundles probably supply at least some adjacent lobules and ducts. Is the initiating cause of PIN related to the blood supply or the ducts? More specifically, are all the acini in a PIN cluster drained by one and the same duct? Are there any acini drained by one duct that do not have PIN, whereas all the other acini drained by the same duct have PIN? Are all the PIN acini that are drained by one duct supplied by the same vascular bundle? Does anyone vascular bundle supply both a PIN cluster and a non-PIN cluster? Are there any PIN clusters whose associated acini originate from only a part of the acini of two or more adjacent ducts? Such observations should help resolve whether the blood supply or the acini/ducts are crucial in the initiation of PIN and PCA. Are the ductal lumenal contents, the prostatic fluid, a primary initiating cause of PIN and PCA? Such fluid contains many chemical compounds which have previously been causally implicated in initiating other cancers. These include androgens, growth factors, prostaglandins, cholesterol epoxides, low density lipoproteins, macrophages, neutrophils and environmental chemical carcinogens. Those lumenal factors that originate from the blood supply (e.g. androgens, some prostaglandins, low density lipoproteins, cholesterol, macrophages, neutrophils and chemical carcinogens) all have all the same difficulties as systemic factors. Prostatic fluid contents do not appear to be a major cause of these lesions because: 1) If they were causal, we would expect to find a prostatic zonal distribution of lesions that reflected the prostatic acini distribution, and this is not the case as evidenced here by the report by Qian et al. The strictly anterior periurethral transition zone and the central zones do not get their fair share of lesions. It is important to note that this is not contradicted in the report by Qian et al. that about 30% of PIN lesions are in the transition zone, which is usually expanded to over one-half the prostate volume by benign prostatic hypertrophy (BPH). 2) Being crucial for reproduction, the acinarlductallumenal contents do not have sufficient variation to account for PIN and PCA because they are always at about the same concentrations in adult men, as evidenced by: (a) Prostatic fluid is continuously produced by the secretory epithelial cells, and even after repeated ejaculations, the essential factors are rapidly replaced. This is necessary to keep man, and many male animals, in a constant state of reproductive readiness for the usually short notice given by the female of her sexual receptiveness for fertilization. (b) Probably the negative pressure created by the relaxing sling muscles of acini after ejaculation sucks some prostatic fluid back into the acini,
International Consultation on PIN - Invited Comments· 927
thus sustaining the exposure of acini to the usual fluid. (c) Probably the closing of the prostatic external urethral sphincter after ejaculation causes retrograde flow of prostatic fluid from the urethra back up into the ducts, just as it causes prostatic fluid ejection into the bladder every time after ejaculation. Such retrograde flow helps keep acini and ducts constantly bathed in prostatic fluid. (d) After ejaculation the sling smooth muscles relax from their pre-ejaculation circumference. This decreases the total acinar volume and thus retains relatively constant concentration of the returned fluid and the area exposed to this fluid after ejaculation. (e) The prostatic fluid is probably only diluted during the relatively short period of sexual excitement just prior to ejaculation when the lumenal cells secrete volume expanders which are quickly reabsorbed if there is no ejaculation. (f) The sling smooth muscles' capacity to contract when stretched beyond a certain limit, provides, through nocturnal emissions, a mechanism to help maintain within relatively narrow limits the amount of prostatic fluid in the acini and ducts at any time. (h) Probably also periodic sexual excitability helps keep this fluid within a narrow range of volume and concentration. Thus, there are multiple mechanisms in adult men and most animals for keeping nearly constant the volume and the concentration of prostatic fluid throughout adult life. This apparently has no universal bad consequences. Most animals have no PIN or PCA, and human PCA varies by a multiple of about fifty between different populations, all exposed to nearly identical prostatic fluid. Thus, I conclude that prostatic fluid and its contents are not sufficient to initiate PIN or PCA. In summary, the above reasoning with empirical disproofs, if correct, establishes that none of the previously suggested causes of PIN and PCA can alone be the main initiating cause of these lesions. It might be suggested at this point that it is necessary also to evaluate all the possible combinations of the above disproved causes of PIN and PCA which were evaluated only in isolation. Generally, this is not the case, because so conjoining independently disproved causes will usually produce only even weaker causal options, because the conjoined disproved causes must then rebut all the relevant combined disproofs. The only way to get around this difficulty of constructing new possible causal explanations out of apparently disproved possible causes is to establish new causal interactions resulting only from the unique conjunction being evaluated. Such consequential causal interactions must nullify not just some, but all the disproofs, not only the disproofs of one conjunct of the new option, but all disproofs of
all conjuncts. Also, any such rebutting causal interaction must be based on new empirical data, not just possibilities, that were not available at the time of the original evaluation and disproof of each conjunct separately; for otherwise, that interaction data would have been an adequate rebut in the original analysis. Thus, establishing possible combinations of disproved causes m~st predict and prove by new research some new relevant interactions. Such attempts at proving new conjunctions is usually much more difficult than merely attempting to patch up only one disproved explanation. Attempts to so construct a better explanation by combining previously disproved explanations are ordinarily justified only if no additional independent explanation is adequate to explain all the available data. Otherwise, inferentially so defining and seeking such new data to support such disproved explanations will have no restrictions and be an endless logical game generating nearly limitless possibilities. Thus, it should be undertaken only if no explanation can explain all the data available. Thus, I will not further analyze possible conjunctions of explanations disproved above because a different and adequate explanation of the data is now available about PIN and PCA (see below). Likely Causes of PIN and PCA Can partially blocked ducts, resulting in increasing inner acini pressure, explain all the observations about PIN and PCA? This possible causal explanation of these lesions arose from work on three prostatic problems, all related to PIN and PCA, and each which must be explained. Specifically, when all possible explanations of these three problems were listed, the explanation that was common involved inner acini fluid flow and pressure dynamics. The three initial problems were: 1) In clusters of PIN lesions, how can we explain the approximately one hundred valleys between adjacent tufts, assuming on average about ten tufts per acinus and about ten acini per cluster? 2) I noted that both in the PZ and in the transition zone (TZ), cysts were frequently closely associated with PIN lesions. What are the detailed mechanisms and stages of the evolution of prostatic cysts? 3) How can we explain the detailed structure of the PIN cribriform pattern? It apparently gradually evolved from tufts, but why did PIN tufts often have so many cell layers between their opposite sides and in the valleys between the lumenal cells and the basal cells? Both areas were usually about ten cells thick, whereas, in the PIN cribriform pattern, these areas were often only two or three cells thick. Its easy to understand PIN hyperplasia adding cells as it might have evolved from the flat pattern to the tufted pattern. However, why diminish the number of these cells as the tufted pattern evolves into the cribriform pattern? How could this be adaptive? As reported at this conference by Bostwick, PIN
928 . H. Harvey
tufted patterns usually outnumbered cribriform patterns. The cribriform pattern appears more complicated than that of the tufted pattern. Thus, it was hard not to believe that the former evolved from the later. How? Cysts must have evolved from a cluster of adjacent acini which had their common duct blocked, at first partially and later, completely. At some crucial ductal pressure of blockage (the "malignant blockage pressure"), the combined force of contracting acinar sling muscles, ductal muscles, capsular muscles, and surrounding skeletal muscle groups could not adequately push the prostatic fluid past the blockage during ejaculation. Consequently, the fluid gradually increased the total volume, and the pressure within the blocked acini and ducts upstream of the blockage continued to rise rapidly, especially if there was continued sexual excitation and ejaculation. This rising pressure within adjacent acini of the same lobule caused an ever tightening vice-like grip on the stromal and epithelial tissue trapped between adjacent acini. This increasing pressure eventually pinches out these trapped cells and tissue, similar to lung emphysema. This breakdown of adjacent acinar walls enlarges the cyst until its eventual boundary is only those walls at the periphery of the cyst which were not subjected to the two-sided vice-like action that pinched out other acinar walls. This cyst formation took place eventually in all acini of a lobule distal to the blockage of the common duct. Before reaching the blockage pressure for cyst formation, there was probably a long preceding adaptive stage to the increasing inner acinar pressure. Probably, such precystic stages exist in the prostate with a high frequency, possibly even more frequently than that of the cysts to which they give rise. Since no morphological forms of such precystic lesions have been identified, I explored the possibility that the entire gradually adaptive process of acini to partial duct blockage was identical to all the successive stages of acinar evolution of PIN patterns. PIN patterns are the pre cystic morphological forms in the prostate. This proved eventually to have enormous explanatory power. Unlike the bladder, a cluster of acini of a partially blocked duct has a very limited capacity to increase its contractive force by smooth muscle hyperplasia or hypertrophy because the acini are usually close together with little stroma between them. Initially, the partially blocked acini enlarge as much as the stroma will allow as a result of increased inner acini pressure. Such enlarged acini have often been a hallmark of PIN identification as reported at this conference by Bostwick et al., Montironi et al., Algaba, and Qian et al. Such enlarged adjacent acini encroach upon one another, further compromising their capacity to enlarge further or to enlarge the sling muscles to adapt to increasing inner acinar pressure. Any small acini in a cluster of PIN are usually boxed in by surrounding acini, or, as explained later, have recently budded off from adjacent larger acini. Acini associated with an adjacent duct that is not partially blocked retain their normal size. As
pressure continues to rise in a partially blocked duct, acinar enlargement will not be adequately adaptive to shift sufficient prostatic fluid past the blockage, possibly because the stretched sling muscles have reached the upper limit of the effective force they can exert against the higher fluid pressure in the acini. At a higher pressure, acini use a different adaptive process of infilling their lumens with hyperplasic cells. First, a few layers of flat PIN are produced; and, if this is not adequate to shift sufficient prostatic fluid past the blockage, then tufts are formed, since these usually result in even more infilling. Since all acini in a PIN cluster are exposed to approximately identical inner pressure, they are all at about the same stage of PIN pattern evolution, as often observed. To help understand why PIN hyperplasia requires the valleys of tufts, imagine a cook's basting syringe as being similar to an acinus and its duct. The cook's hand, like a sling muscle, squeezes hard on the rubber bulb of the syringe, creating inner bending and creasing of the bulb as it is forced to contract and expell the fluid. Similarly, at ejaculation in an acinus, the valleys between tufts function like the creases in the bulb, allowing for maximally efficient (i.e. complete and rapid) evacuation with minimum constriction and hyperplasia. These valleys of tufts also decrease the flow resistance and turbulence of the rapidly outflowing prostatic fluid at ejaculation when the contents must be emptied within the relatively constant and short contraction time of the sling muscles. Thus, a tufted PIN pattern may be an indication of a more sexually active man than a thick flat PIN pattern. Each hyperplastic acinus in a cluster of PIN acini contributes additively to help overcome the downstream blockage. Further, ducts themselves, in contrast to acini, are rarely similarly hyperplastic, because such would be counterproductive in decreasing the total outflow volume in the relatively constant contraction time of ejaculation. The process of producing flat and tufted PIN patterns requires the increased proliferation capacity produced by high grade PIN (HGPIN) cells. If these patterns of infilling acini are adequately adaptive, the proliferative cells of HGPIN decrease and we see the flat and tufted patterns of low grade PIN (LGPIN). These LGPIN patterns must be retained indefinitely if the cause of the blocked duct does not abate. If the blocking pressure continues to increase on the draining duct, at some point the infilling will no longer be adequately adaptive. Suppose that this resulting increased inner acini pressure is still below that "malignant blockage pressure" necessary for the formation of cysts. At a slightly higher inner acini pressure, tufts can no longer retain their structure. The increased pressure on the opposite sides of a tuft squeezes the sides inwards and into the micropapillary pattern. The same increased inner pressure on the valleys of the tufts reduce their thickness from about ten cells to two or three cells. Fewer lumenal cells are columnar, and more become cuboidal at these increased pressures. All these decreases in inner lumenal cells, as they succumb to in-
International Consultation on PIN - Invited Comments . 929
creased pressure, proves that infilling is no longer effectual. Consequentially, acini adapt another strategy to survive in this environment. They must subdivide to conquer. Tufts become micropapillary vine-like growths which join up as bridges, arches, trabecular bars, or cribriform acinP. All of these forms have the capacity to subdivide the lumen via ingrowths of basal cells and fibrovascular tissue. Thereafter, the epithelial cells grow new, confluent, cul-de-sac, partial spheres which subdivide the original lumen into multiple smaller lumens, each surrounded by a complete stroma, including a functional sling muscle. These new acini cumulatively have a contractile force that is sufficient to again shift the prostatic fluid past the blockage at the higher inner pressures. The initially thin forms of the micropapillary pattern are necessary if they are to grow into such a lumen with such high pressure and fast flowing turbulent fluid . This environment is similar to that at the bottom of a deep, fast flowing river where only thin plants can grow far out into the fluid. Such prostatic micropapillary patterns will minimize turbulence and maximize the flow rates at ejaculation. Though necessarily thin, these micropapillary and cribriform stalks are rarely less than two cells thick because only in this way can they retain normal intercellular functions which will later prove necessary for their adaptive survival. In any functional unit of micropapillary or cribriform stalks, two or more cellular units cooperate in presenting three cellular surfaces, each with a different function. The lumenal surface, usually cuboidal, can later differentiate into secretory columar cells. The inner surface, now back-to-back, is programmed to make adhesions via integrins with basal cells and thereby with the fibrovascular core that will eventually grow into the center of the extended micropapillary pattern. This fibrovascular core will supply the necessary nutrients as well as provide the stromal basis necessary to subdivide an acinus into two or more smaller acini, each capable of contraction. If each acinus of a PIN cluster similarly subdivides into smaller acini, then the combined contractual force is greatly increased, resulting in shifting more prostatic fluid beyond the ductal blockage at ejaculation. The third cell surface of these growing micropapillary and cribriform stalks is that surface which adheres to the neighboring identical cells. These surfaces will make possible the confluent saccular growth necessary to form the subdividing acini. Micropapillae often join in a tip-to-tip pattern. Adjacent papillae join tips to form bridges and arches, and opposite papillae join tips to form trabecular bars. Combinations of these forms often lead to the cribriform pattern. Some cribriform acini often have joined micropapillae forming a ladder of curved bridges with all the rungs pointing downstream. This is because the downstream flow of ejaculated prostatic fluid forced adjacent micropapillae to bend downstream where their tips touched and joined to form the rungs of such a ladder. These resulting new scaffold-like structures in-
side an acinus are stronger, grow thicker, and allow congruent growth into new saccular acini, each with its own stroma. Small acini, and large acini with narrow ends (e.g., a budding area at the blind end of a large acinus or the narrow end where an acinus joins its draining duct) will often have no micropapillary ingrowth. This is possibly because such small areas on subdividing could only form very small and ineffective new acini. Possibly, the preferential formation of such optimally large new acini on subdivision is the reason why so many new short micropapillary sprouts appear to bisect the remaining acinus. The sequential evolution of the PIN patterns helps to explain why, when we see mixed forms of these patterns in one acinus or cluster, the patterns of mixture are usually those of adjacent evolutionary patterns, suggesting that they do not occur randomly, but evolve in an orderly fashion from one pattern to another, possibly due to increasing inner acini pressure. The mechanism of progressive adaptive responses of acini to partial ductal blockage can explain the forces that shape micropapillary and cribriform patterns, and also the patterns seen in clear cell cribriform hyperplasia, cribriform intraductal carcinoma, cribriform carcinoma, and cribriform basal cell hyperplasia. Thus, possibly all these lesions are caused by blocked ducts, especially since most have enlarged acinar clusters as well as a cribriform pattern. How can an adaptive mechanism of PIN evolution lead in some cases to PCA? Let us define a high grade PIN (HGPIN) cell as that cell seen frequently in HGPIN which has an enlarged nucleus, an enlarged prominent nucleolus, peripheral nuclear chromatin, is stratified, often with nuclear crowding, is not strictly a basal cell (which is androgen receptor negative, bc1-2 positive, PSA negative, located at the periphery of the acinus and produces some factors of the basement membrane) , nor, though at times located at the lumenal surface, is not strictly a secretory cell (which is androgen receptor positive, bc1-2 negative, PSA positive, at the lumenal surface, usually columnar with a basal nucleus). At times HGPIN cells appear to have some of the properties of the basal cells from which they probably arose, and also some of the properties of the secretory cells into which they can differentiate. Yet, the HGPIN cell does not appear to be merely a cell in transitional development or differentiation. The most common cell in HGPIN, it appears to be the primary cause of hyperplasia in HGPIN, since rare mitotic figures are usually found within the HGPIN cell layers. This was reported by Helpap, who showed that HGPIN cells have a proliferative rate similar to those of PCA, and both have a higher proliferative rate than normal cells or BPH. By definition, the HGPIN cell is less frequently found in low grade PIN (LGPIN) which probably indicates that LGPIN has reached a point of adaptive adequacy (i.e., sufficient hyperplastic infilling and/or acinar subdivision to overcome the blockage), and that it no longer needs the proliferation of HGPIN cells. The report
930 . H. Harvey
by Montironi et al showed that plump basal cells were observed more in LGPIN than in HGPIN suggesting that these plump cells are under less luminal pressure than cells of HGPIN. Silvestri et al. reported that there is little additional PIN in men over 65 years old. This may be due to the decreased ejaculation that accompanies ageing, and this results in less increased acini pressure; thus, no further PIN adaptive growth is necessary. To this end, many HGPIN cells lose their proliferative capacity, possibly by terminal differentiation and become LGPIN cells. However, the final adaptive PIN pattern attained during hyperplasia is probably retained indefinitely. Such indefinite retention is also commonly seen in many other tissues e.g., the endothelium in atherosclerosis, blood vessels with high blood pressure, heart muscle with decompensation, and skin with increased friction. The presence of many HGPIN cells probably indicates that the acini are still adapting to ductal blockage and increased inner acini pressure by proliferating, resulting in acinar infilling and/or subdividing. HGPIN cells are remodelling their acini, and retain the capacity for proliferation and mobility. Such mobility requires a cell membrane that has lost many of its intercellular adhesions. Thus, HGPIN cells are stratified, with poorly defined cellular boundaries and confluent growth resulting in nuclear crowding. As in other precancerous lesions, the proliferative compartment has moved from the basal cell layer, possibly in part to provide a protective buffer layer of the genetic material in the basal cells during the ensuing period of relatively rapid proliferation. Probably more important, layers of HGPIN cells make possible proliferation and mobility responses to the focal demands of the environment. Microscopically, the HGPIN cell is indistinguishable from the PCA cell. This cell is probably rarely genetically mutated in any way relevant to causing PCA since it arises independently and often simultaneously in clusters, for example, at the base of each of about ten tufts per acinus and in each of about ten acini in the average cluster of HGPIN. Further, repeated efforts to date have found no consistent pattern of gene mutations causing PCA, whereas nearly constant patterns of mutation have been found in a number of other cancers. It is therefore reasonable to ask if the nonmutational mechanism of adaption to ductal blockage can explain the evolution of PIN to PCA. Suppose that an acinus with HGPIN suffered an acute increase of acinar pressure due to ductal blockage and/or frequently repeated sexual excitation without ejaculation. It is imaginable that, with such an acute increase in prostatic fluid pressure, the basal cells and the basement membrane would disrupt. The finding of Bostwick et al. l shows that basal cell disruption appears equally frequently in all patterns of PIN, and supports this possibility that the process causing such disruption could be independent of any particular PIN pattern. In such a disruption, basal cells and lumenal cells are unlikely to be propelled through the disruption and into the stroma because they are so strongly adher-
ent to neighboring cells and surrounding structures (e.g., to the basement membrane). Even if propelled into the stroma, these cells are unlikely to proliferate since their proliferative capacity is contingent on integrin and E-cadherin attachments. These cells are programmed for confluent growth only in a specific environment. Also, these cells would have mobility limited by intercellular attachments. The basal cells and luminal cells probably react like most other adherent and confluent epithelial cells throughout the body, and simply heal the disruption. Conversely, HGPIN cells could easily be propelled through such an acinar disruption. These HGPIN cells are programmed to proliferate and move. Furthermore, if the high pressure environment of the acinus is removed from the HGPIN cells when they are propelled into the stroma, they probably are programmed in this environment to differentiate into lumenal cells with congruent growth, cell adhesions, and secretory capacity. Such properties of HGPIN cells in the stroma are sufficient to make them identical to the cells of PCA. In the stroma, one HGPIN cell could produce all the sequelae of PCA. First, it could proliferate and partially differentiate along the pathway of a luminal cell into a more adhesive cell which could grow into the congruent, single cell layered sphere of an acinus of well- differentiated PCA. Due to the confines of the stroma these new acini would often be relatively small. Since this HGPIN cell in the stroma is not terminally differentiated, it still retains some of the properties of basal cells. Thus, it could produce a basal lamina around all invasive cells and acini, as reported at this conference by Nagle et al. The single small acinar sphere produced by a HGPIN cell proliferating in the stroma would have no draining duct. Thus, its constitutive secretions into this new stromal acinus would rapidly raise the inner pressure. Consequently, this first stromal acinus would soon disrupt. Such disruption would be at a relatively low inner pressure, since this acinus has no protective reinforcing layers of basal cells, sling muscle, or most normal extracellular matrix factors. Also, the raised inner acinar pressure may cause the lumenal HGPIN cells to revert to their prior condition of decreased intercellular adhesions. This could facilitate such further disruption. This disruption could initiate the growth of an adjacent stromal acinus, possibly from the disrupted pointed end of the parent acinus. Possibly, the enzymes that broke down neighboring cell adhesions to allow for proliferation and mobility, when a HGPIN cell was still within a PIN acinus, are used in the stroma for progressive invasion and malignant growth. Such altered enzymes of precancer and cancer cells was reported here by Pretlow et al. . Cardillo reported an altered enzyme might be cathepin D, and Lucarini reported it might be 72 kDa metalloproteinase. In the stroma, the invasion would probably follow the path of least resistance. Those neoplastic cells that found themselves trapped between layers of smooth muscle may not have space to form rounded acini, and would grow as chains or sheets of invasive cells.
International Consultation on PIN - Invited Comments . 931
The process of forming new small acini in the stroma could repeat itself indefinitely via repeated stages of secretion, disruption, and regrowth. Thus, all forms of invasive PCA could arise from one HGPIN cell propelled into the stroma. Further, it appears that all forms of PIN and PCA can be explained without necessary recourse to gene mutations. Thus, it is possible that gene mutations are neither necessary nor sufficient to explain PIN and early PCA. When they occur, nearly randomly, they may merely accelerate or slow down the process. What initially blocks the prostatic ducts, resulting in the gradually increasing inner acinal fluid pressure which ultimately causes PIN and PCA? Pressures external to the prostate, from a full bladder, increased fat tissue, hard bulky feces in the rectum, or a tightly constricted penis during ejaculation, are all probably of little causal significance here because (1) their forces are all fairly evenly distributed throughout the prostatic tissue; (2) they are usually transitory, and (3) their global effects on the prostate cannot explain the patchy and focal distribution of most PIN and PCA lesions. Sources of duct blockage from within the body of the prostate include the consequences of (1) Inflammation, stones, or proteineous plugs; though properly located in the larger ducts, these are too rarely associated with PIN and PCA which occur in 50 % to 75% of all men; (2) infarction or venus blockage; though associated with ageing, these are usually located too peripherally to block the drainage ducts, and are also rarely associated with PIN or PCA; (3) hyperplasia of ductal epithelium (in contrast to hyperplasia of acinar epithelium) due to increased testosterone, estrogens, or selected chemical carcinogens; though these have been undoubtedly the cause of PCA in some species of lower animals, ductal hyperplasia is too rare in man to be a major cause of ductal blockage to explain a lesion that occurs so frequently in man; and (4) BPH.
BPH as the Cause of PIN and Cancer? BPH may be the primary cause of nearly all PIN and PCA in man. First, BPH increases with ageing with a frequency that is greater than that of PIN or PCA. Such a relative frequency is an attribute of a necessary, but not sufficient, cause of a pathologic lesion. Such a prelesion must occur more frequently than its resulting pathology so when the other necessary causal conditions are not also satisfied there can be cases of the prelesion which do not result in the pathology. In other tissues, this condition is nearly universally true: the prelesion atherosclerosis occurs much more frequently than its resulting lesion, infarction, both with regard to the number of persons with the prelesion, as well as the number of pre lesions in each person. Similarly, there are many more persons with immunological evidence of tuberculosis infection than with tuberculosis pathologic lesions. Further, there are more large intestinal adenomas than there are large adenocarcinomas;
more skin pre-cancer lesions than skin cancers; and more breast pre-cancer lesions than breast cancers. Part of the evidence that PIN is a prelesion of PCA is its greater frequency. Second, BPH is the only prostatic lesion that occurs sufficiently frequently to satisfy this frequency condition with regard to PIN and PCA. Third, the slowly gradually expanding nodules of BPH represent the kind of chronic causal mechanism necessary to explain the long pathological evolutionary process, beginning with the slightly enlarged acinus, evolving over years through all the patterns and stages of PIN and resulting in cysts or PCA. Fourth, the enormous growth and hydraulic forces of BPH are strong forces needed to block the mature and active ducts of the PZ. Such forces can push into the multi-layered muscular wall of the bladder, and compress the nearly entire PZ, often to one-tenth of its original volume. Finally, BPH is usually present in prostates that have PIN or PCA, as reported at this conference by Qian et al., and Silvestri et al. Previous research about this association has had inconsistent results, most being negative, because (1) such research relied on a measure of correlation that would not capture the uniqueness of this relationship; if BPH is a necessary part of a more complex sufficient cause (i.e. there are other necessary conditions), we can expect to find many cases of BPH that are not associated with PIN or PCA, but we should find almost no cases of PIN or PCA which are not associated with BPH. Disproof of the association would be finding a significant number of cases of PIN or PCA without associated BPH. Even 50% of cases of BPH with no associated PIN or PCA is irrelevant as disproof; and (2) such research often attempted to establish a close spatial relationship between BPH and PIN/PCA as has been found for PIN and PCA. However, the causal mechanisms are different. Thus, in testing for BPH, allowance must be made for causal action-at-a-distance if the mechanism is that of blocking an outflow drainage duct near the urethra that causes PIN and PCA in the acini deep within the PZ. In fact, those acini closest to the point of blockage are likely to be affected least since their fluids are the first to squeeze by the blockage, and any retrograde flow of fluid is more likely to concentrate in the larger cluster of distal acini which will produce the greatest negative pressure on relaxing after ejaculation. Those distal acini are also more likely to produce the greatest combined expulsive force during contraction, and thus are more likely to disrupt into the stroma. Thus, not finding a close spatial relationship between BPH and PIN or PCA is expected if BPH is the main cause of PIN and PCA. Causes of BPH I will outline the essentials (with no disproofs) of what may be the cause of BPH, because (1) in this hypothesis, BPH is the primary cause of PIN and PCA,
932 . H. Harvey
and (2) understanding the causal processes of BPH will suggest solutions to many of the difficult and significant problems of how PIN and PCA are related to atypical adenomatous hyperplasia-adenosis (AAHadenosis) which usually arises in BPH. First, in the early development of the prostate, growth of the ducts and acini were prevented from overgrowing each other and nearby structures, especially the inner periurethral muscle, by growth inhibitors, especially transforming growth factor beta (TGF p). Second, during ageing, these growth inhibitors decrease due to: (1) General ischemic atrophy in the PZ due to atherosclerosis of the capsular branches of the prostatic artery whose endothelium has been repeatedly damaged as it pumped at greatly increased blood pressure against the contracted capsule and acini during ejaculation. Thus, atherosclerosis arises in these arteries and the coronary arteries for the same reasons; (2) Decreased ejaculation results in disuse atrophy of the inner periurethral muscle (IPUM) but not of the sling smooth muscles, which are kept sufficiently active by their intrinsic stretch-contract properties, resulting in nocturnal emissions; (3) Decreased exercise of abdominal muscles results in disuse atrophy of these muscles and of the anterior prostatic fibromuscular zone which grew large in man to control his increased pelvic and abdomenal pressures resulting from his taking an upright stance; (4) Loss of acini throughout the prostate, but mainly in the PZ, due to cyst formation, infarction, infection, stones, etc. Third, decreased growth inhibitors result in replacement growth via stromal and epithelial cells producing (a) prostatic tissue growth factors, such as hepatocyte growth factor (HGF), and (b) local paracrine growth factors, such as epithelial growth factor (EGF) and transforming growth factor alpha (TGF a), and their receptors. These growth factors cause prostatic gland growth, usually patchy in the PZ, and much more extensive in the TZ, especially near the IPUM. Fourth, most replacement growth in the mature adult prostate takes the abnormal form of epithelial cells growing into an acinus or duct lumen, rather than, as in earlier development, into the surrounding stroma. Such abnormal growth into a duct is removed from the normal stromal growth inhibitors, so this growth often tontinues indefinitely. This explains why at times one sees epithial growth only from the inner-nodule side of a duct (the outer side is growth-inhibited)2. Fifth, such inward growing epithelial cells produce two sets of stromal growth factors: (1) One set stimulates the inward growth of fi brovascular cores which provide nutrients to sustain the growing epithelial cells. These cores can subdivide acini into multiple independent congruent nodules; and (2) The other set is produced when the hyperplastic epithelium begins to produce small acini within the duct or acinus. These growth factors function to stimulate the mature stroma to initiate new coordinated
growth of the extracellular matrix, but there is no mature stroma inside the ducts of the abnormally inwardly growing epithelial cells. Thus, these unused stromal growth factors gradually diffuse outward towards the periphery of the expanding new growth .. In the periphery, these growth factors find their targets and stimulate a stromal response that results in excessive fibromuscular tissue surrounding a nodule of BPH or AAH-adenosis. Sixth, the epithelium of nodular growths produces only constitative amounts of secretions, and most of their ducts are ineffectual because: (1) The sling muscles and their nerves do not mature; (2) The large fibromuscular nodular periphery blocks them; and (3) Especially in the TZ, the ducts have been stretched a great distance from their origin near the vermontanum; thus, in these ducts, secretions gradually accumulate to form small cysts, since inner glandular pressures are so much lower here than in PIN. Epithelium at the edge of a nodule matures more than in the center, since it is near mature stroma and produces complete extracellular matrix and more secretory epithelium, and thus experiences greater inner glandular pressure and may produce PIN and PCA. Also, large cysts can thus form at the periphery of a nodule. Seventh, purely stromal nodules result, usually close to the urethra, if stromal growth factors stimulated by HGF, EGF and TGF a , are unable to attract the local inward growth of ducts which is necessary to produce the stromal growth factors. Eighth, AAH-adenosis results from the maturation of BPH wherein the tightly packed epithelium form small acini if there are no local growth inhibitors. Only small islands of AAH-adenosis develop in the PZ where the more mature tissues produce more growth inhibitors. AAH-adenosis in the PZ has low inner glandular pressure for the reasons given above; thus, these glands rarely produce PIN or PCA lesions in the PZ. Similarly, there are rarely basal cell disruptions in AAH-adenosis, probably because without mature individual sling muscles, acini cannot at ejaculation attain the increased pressures necessary to produce such disruptions. However, with immature extracellular matrix (including abnormal sling muscles), AAH-adenosis in both the TZ and the PZ can break down the neighboring weak interacinar walls and eliminate some of the basal cells. Any acinus at the periphery of AAH-adenosis may have a mature extra cellular matrix, and thus is capable of producing PIN, HGPIN cells, disruption, and PCA since those acini can attain the necessary high inner gland pressures to produce these lesions. This explains why Epstein reported that finding convoluted acini near disrupted basal cells in AAH-adenosis may help predict eventual PCA, since such convolutions indicate higher pressures necessary to produce PIN and PCA.
International Consultation on PIN - Invited Comments· 933
Application of the Hypothesis Using the combined causal mechanism of both PIN/ PCA and BPH/AAH adenosis, we can attempt answers to the following perplexing questions which have been raised at this conference. When is AAH-adenosis Cancer?
If only one HGPIN cell in the stroma is necessary and
sufficient to give rise to all the forms of invasive PCA, then subsets ofHGPIN cells can be analyzed as follows: (1) If no HGPIN is in AAH-adenosis (or PIN), then the adaptive process is probably no longer active and the level of pressure at which adequate adaptation was achieved is reflected in the resulting PIN pattern. This PIN pattern has no prognostic significance if there are no HGPIN cells present. In all prostatic zones, there are probably many instances of benign disruption of the basal cells without PCA because there are no HGPIN cells present to be propelled into the stroma. Such benign disruptions probably occur whenever exceptionally high ejaculatory pressures are attained. These disruptions are quickly healed by epithelial cell growth. (2) If HGPIN cells are the most common cell in AAH-adenosis (or PIN), then the adaptive process is probably active. HGPIN cells are proliferating and are loosely adherent, and the high acinar pressures are in danger, especially at ejaculation, of disrupting the basal cell and basement membrane and projecting HGPIN cells into the stroma and causing PCA. (3) If there are only a few HGPIN cells present, a judgement must be made as to whether adaption has been achieved, and thus further hyperplastic proliferation is unlikely, because we are witnessing the retirement of HGPIN cells. On the other hand, we may be witnessing the beginning of a hyperplasic proliferative response in which the number of HGPIN cells increases. Noting the PIN pattern may be helpful. Flat and tufting patterns are often a sign of the adaptive process. On the other hand, a micropapiUary pattern would indicate greater acinar pressure and the continued attempt at adaptation. New acinus formation may indicate adaptive mastery of the increased pressure. However, if BPH is the primary cause of ductal blockage, the entire process is likely to be relentlessly progressive until BPH is halted. Can Biopsy Disperse Even one HGPIN Cell into the Stroma and thus Cause Invasive PCA? Yes, on my assumptions, regretfully. Such a consequence is another possible explanation as to why repeated biopsy of HGPIN often finds PCA. It should be possible to test this possibility on animals transplanted with human PCA. Some evidence can be gained by merely noting the results of repeated biopsy of dif-
ferent prostate areas. This remote possibility that biopsy itself could cause PCA suggests further the importance of the work reported at this conference by Bologna et al. with tissue culture techniques with massage expressed prostatic cells to screen or diagnose PCA or PIN.
What is the Source of Prostatic Cells Obtained by Massage? We cannot believe that massage pushes cancer cell from the stroma into the acinus, through the duct, and out the urethra. Also, if that were the case, prostatic massage itself alone could probably disperse cancer cells through the stroma. The cultured prostatic cell is probably the HGPIN cell. This is consistent with the suggestion that the HGPIN cell displaced into the stroma is sufficient to produce PCA without any necessary mutation.
Why Does Adenocarcinoma so Rarely Have the PIN Pattern? The hyperplasic cells of PIN are probably regulated by basal cells which are not present in PCA. It is also possible that in the stroma the loosely adhesive HGPIN cells form cancer acini which disrupt at inner pressures that are too low to initiate PIN.
Why Does Clear Cell Cribriform Hyperplasia (CCCH) Rarely Have HGPIN Cells? Without HGPIN cells, CCCH is unlikely to produce PCA. CCCH is almost always in the TZ, usually associated with BPH. As in other BPH growths in the TZ, CCCH probably has lower acinar pressure, because (1) it has a scant fibromuscular stroma that probably cannot contract sufficiently to raise inner acinar pressure significantly, and (2) its clear cytoplasm, cuboidal cells, and lack of mucin and acid mucin probably indicate that its secretory capacity is not fully differentiated. Both of these facts suggest that CCCH cannot raise acinar pressure sufficiently to cause PIN. CCCH is also interesting because the existence of the cribriform pattern without HGPIN suggests that this may merely indicate a chronic adaptive process, regardless of the inner acinar pressure levels. Finally, the above mechanisms of increased acinar pressure may also explain the initiating causes of cancer of the breast, pancreas, salivary glands, and endometrium, all of which, in common with prostatic cancer, are associated with inner ductal growths, cribriform patterns, cyst formation, lack of consistent gene mutations, and uncertain initiating mechanisms.
934 . H. Harvey
2 McNeal, John (1990) Pathology of Benign Prostatic Hyperplasia. Urol Clin of North Am 17: 477-486
1 Bostwick DG, Mahul BA, Dundore P, Marsh W, Schultz DS. (1993) Architectural Patterns of High-Grade Prostatic Intraepithelial Neoplasia. Hum Pathol24: 298-310
Received June 14, 1995 . Accepted July 17, 1995
Key words: Prostatic cancer - Prostatic intraepithelial neoplasia - Benign prostatic hyperplasia - Atypical adenoma taus hyperplasia - Initiating causes H. Hale Harvey MD, PhD, MPH, Puckaster Close, Puckaster Lane, Niton Undercliff, Ventnor, Isle of Wight, England, P038 2 LZ, Telephone (01983) 730206