Академический Документы
Профессиональный Документы
Культура Документы
ABSTRACT The relationship between periosteal new individual disease states. It was established that disease
bone formation and a number of infectious and metabolic progression, rather than disease type, was the most im-
conditions frequently seen in archeological human skele- portant determinant of periosteal lesion appearance. A
tal remains was investigated by studying human long critical analysis of the bioarcheology literature pertain-
bones demonstrating periosteal new bone formation ing to the recording and interpretation of periosteal reac-
archived in two London, UK, pathology museums: the tions determined that the varied pathogenesis of perios-
St. George’s Hospital Pathology Museum and the Hunt- teal new bone formation has been largely ignored in
erian Museum. The samples were subjected to macro- favor of a diagnosis of ‘‘nonspecific infection.’’ Assump-
scopic and radiographic analysis to determine if the tions regarding the infectious etiology of periosteal
characteristics of their periosteal lesions were specific to lesions have become embedded into the bioarcheology lit-
the corresponding disease states. The results demon- erature potentially skewing the results of skeletal popu-
strated that no qualitative or quantitative characteristics lation-based paleoepidemiological studies. Am J Phys
of the periosteal reactions emerged that were specific to Anthropol 137:48–59, 2008. V 2008 Wiley-Liss, Inc.
C
Proliferative periosteal reactions, or ‘‘periostitis,’’ as disease, assuming that the same pathological changes in
they are often termed, are a common occurrence in ar- skeletal remains used in the diagnosis of a clinical
cheological skeletal remains, being frequently seen on patient can be applied as diagnostic criteria in the inter-
the long bones, especially the tibiae. This periosteal new pretation of ancient material (Klepinger, 1983). Thus, in
bone formation, occurring as a response to extrinsic or order to characterize how different pathological condi-
intrinsic pathological factors, initially has the character- tions manifest periosteal new bone formation, diagnosed
istic appearance of woven bone, and remodels over time diseased bones exhibiting this feature must be examined.
into lamellar bone. In spite of its ubiquitous nature, very If the characteristics of the periosteal new bone can be
little investigation of periosteal reactions and how they correlated with a particular pathological condition, it fol-
relate to specific pathological conditions has occurred. lows that the techniques could be applied to archeologi-
Bioarcheologists have most commonly interpreted peri- cal bones, the correlated characteristics could be identi-
osteal new bone formation as evidence for nonspecific in- fied, and a diagnosis might be made. Pathology muse-
fectious disease in past populations, with recent syn- ums are prime repositories of skeletal reference material
thetic works (Larsen, 1997; Steckel and Rose, 2002) of this type, offering specimens with a wide range of
continuing the trend that was popularized by Cohen and associated pathological conditions and, as such, skeletal
Armelagos, 1984 influential volume, Paleopathology at material from two British pathology museums, the
the Origins of Agriculture. Additionally, periosteal new Hunterian Museum at the Royal College of Surgeons of
bone formation also plays a central role in the ‘stress-in- England and St. George’s Hospital Pathology Museum
dicator hypothesis’ (Goodman et al., 1988), within which were examined.
periosteal reactions are again deemed to be a sign of in-
fectious disease. This interpretation of periosteal reac-
tions has continued despite a wealth of clinical literature MECHANISMS OF PERIOSTEAL NEW
(e.g. Resnick, 1995; Adler, 2000) indicating multiple BONE PRODUCTION
pathological etiologies for proliferative periosteal reac-
tions, and has resulted in generalizations made about Periostitis is a poor word for periosteal new bone pro-
pathogen load models in archaeological populations. duction because by definition it assumes that inflamma-
The purpose of this paper is to present the results of
recent research investigating the relationship between
*Correspondence to: Darlene A. Weston, Department of Human
periosteal new bone formation and a number of infec- Evolution, Max Planck Institute for Evolutionary Anthropology,
tious and metabolic pathological conditions; testing Deutscher Platz 6, 04103 Leipzig, Germany.
whether periosteal new bone formation manifests differ- E-mail: weston@eva.mpg.de
ently depending on its etiology. To achieve these ends,
macerated dry long bones curated in pathology museums Received 2 August 2007; accepted 25 January 2008
with records relating to known pathological conditions
were analyzed macroscopically and radiographically. Bio- DOI 10.1002/ajpa.20839
archaeology employs biomedical clinical analogy in the Published online 8 April 2008 in Wiley InterScience
recording and interpretation of the ancient evidence for (www.interscience.wiley.com).
C 2008
V WILEY-LISS, INC.
SPECIFICITY OF PERIOSTEAL REACTIONS IN PATHOLOGY MUSEUM SPECIMENS 49
tion has occurred (Ragsdale, 1993). However, the mecha- a more complicated process due to its longer duration,
nisms involved can be much more complex, as almost which allows for a wide variety of patterns of inflamma-
anything that breaks, tears, stretches, or even touches tory mechanisms to come into play. In bone, vasodilata-
the periosteum (the membrane of connective tissue that tion through hyperemia results in hyperoxia (an increase
surrounds all bones, except on their articular surfaces) in oxygen), which stimulates osteoclast function. Fluid
can stimulate it into initiating bone formation exudation results in edema, an excess accumulation of
(Richardson, 2001). The physical elevation of the perios- fluid, which causes hypoxia (a decrease in oxygen),
teum is often cited as a requirement for the initiation of which in turn stimulates osteoblast function. These dif-
periosteal new bone production. Physical elevation may fering and sequential oxygen states are crucial in the os-
be present, but it is not a prerequisite for a reaction, seous changes of bone inflammation (Roberts and Man-
meaning other factors must be operative (Ragsdale, chester, 1997).
1993). Likely choices include mechanical adaptation or a If the inflammatory process is successful, healing
compensation for weakness secondary to underlying begins. The inflammatory process inevitably results in a
bone resorption, attempts at tumor containment, altered mass of dead tissue and cells, and healing must occur to
circulation, and perhaps bone inductive products ema- replace this matter via the scavenging of dead material,
nating from tumors (Ragsdale et al., 1981; Ragsdale, the regeneration of lost tissue, and repair (Mitchinson
1993). et al., 1996). In bone, scavenging is performed by macro-
Lesions that are caused by the elevation of the fibrous phages and osteoclasts. Repair of the bone tissue
outer layer of the periosteum are formed after the com- involves the formation of granulation tissue, a type of
pression and stretching of blood vessels by agents such loose connective tissue abundant in capillary buds, osteo-
as blood, pus, granulation tissue, neoplasm or trauma blasts, macrophages, and other inflammatory cells. The
(Bush, 1989). This may result in bleeding beneath the granulation tissue creates organization out of a disorgan-
periosteum, with a subsequent reduced blood supply to ized mess, and once its job is complete, the tissue
the bone. If this situation is of sufficient duration, the becomes less vascular and less cellular, with the inflam-
periosteal bone tissue will die. Amelioration of the blood matory cells disappearing, the osteoblasts turning into
vessels will restore osteoblastic activity, producing new osteocytes, and the tissue fluid reabsorbing into the
subperiosteal bone that is deposited on the normal corti- shrinking vasculature. The granulation tissue becomes
cal bone surface (Jaffe, 1972; Bush, 1989). A row of the periosteal new bone, initially composed of woven
refilled Howship’s lacunae forming a reversal line can bone and gradually converted into lamellar bone through
usually be seen on the original cortical surface beneath remodeling.
the newly added bone, testifying to an initial resorptive
phase probably due to active hyperemia (increased blood
flow) (Ragsdale, 1993). PERIOSTEAL REACTIONS IN PAST STUDIES
The modern clinical literature
Inflammation vs. infection The modern clinical literature pertaining to periosteal
new bone production as a whole provides little or no
Inflammation has an important role to play in the eti-
description of the phenomenon itself. For the most part,
ology of periosteal new bone production. Although as
it is merely stated that ‘‘periostitis’’ is present as part of
previously stated, it is not always present when a perios-
the manifestations of the clinical disease. There are few
teal reaction arises, it frequently occurs as a general
descriptions of specific locations or types of periosteal
response to abnormal stimuli such as trauma, neoplastic
reactions, nor are there any gradings of periosteal lesion
disease, or infectious agents (Ortner, 2003), and thus
severity. Some of these difficulties found within the mod-
stimulates periosteal reactions to occur. It is a frequent
ern clinical literature are based on the fact that the ma-
practice in studies of archaeological skeletons to attrib-
jority of researchers rely on radiographs or other types
ute the majority of periosteal lesions to infection,
of imaging to report on and diagnose the pathological
perhaps due to confusion between the definitions of
conditions. Unfortunately, periosteal new bone in its ear-
inflammation and infection (i.e. Lallo, 1973; Mensforth
liest stages does not always appear evident on radio-
et al., 1978; Littleton, 1998). Inflammation is the body’s
graphs and other imaging media and is likely to be over-
vascular response to tissue damage from a variety of
looked in autopsies (Kelley, 1989). Regardless, most peri-
causes, while infection occurs when the body encounters
osteal reactions are part of the expression of a specific,
pathogenic organisms, usually bacteria. Part of the
identifiable disease process (Ortner, 2003) (see Table 1)
body’s reaction is to mount an inflammatory response to
that spans a varied range of conditions, falling under
neutralize the organism and repair or heal damage.
the categories of trauma, joint, infectious, metabolic, vas-
Although infections are among the most common causes
cular and neoplastic disease.
of the inflammatory response, not all such responses are
caused by infection (Bush, 1989).
The inflammatory process occurs no matter what the The bioarcheology literature
cause of the injury, and is initially acute in nature, last-
ing only a few days. The initial inflammatory response, Because of its pervasiveness, periosteal new bone pro-
termed acute inflammation, follows the same predictable duction is a widely recorded feature of archaeological
course: widening of the blood vessels (vasodilatation), skeletons, often being used to assess infectious disease
fluid exudation, and phagocyte recruitment. These proc- at the population level. For example, across a sample of
esses are all localized at the site of injury and occur British skeletal populations, prevalence rates for tibial
approximately within one hour of the injury occurring ‘‘periostitis’’ have been recorded as ranging from 5.9% to
(Mitchinson et al., 1996). If the causative agent persists 8% (Coughlan and Holst, 2000) to 64% (Holst et al.,
(more than about 2 weeks), the inflammatory process 2001) (see Table 2). It is often difficult to assess the
becomes prolonged and is termed chronic inflammation, number of skeletons with tibial periosteal reactions in
TABLE 2. Prevalence rates for tibial periosteal reactions in a sample of British skeletal populations
% of population
Site Period affected Authors Context
Kempston, Bedfordshire Romano-British (AD 43-410) 25% Boylston and Roberts, 1996 Rural cemetery
Eccles, Kent Anglo-Saxon (AD 410-1066) 22–25% Boocock et al., 1995 Rural cemetery
Towton, North Yorkshire Medieval (AD 1461) 5.9–8% Coughlan and Holst, 2000 Mass battlefield grave
St Helen-on-the-Walls, York Medieval (c. AD 1100-1550) 22.4% Grauer, 1993 Urban cemetery
St Andrew’s Fishergate, York Medieval (c. AD 1200-1538) 23% Stroud and Kemp, 1993 Monastic cemetery
Hospital of St James and Medieval (c. 12th-16th C) 39% Lee, 2001 Hospital cemetery
St Mary Magdalene,
Chichester, West Sussex
Hull Magistrates’ Court, Medieval (c. AD 1300-1450) 64% Holst et al., 2001 Monastic cemetery
Kingston-upon-Hull
published reports and articles due to differences in re- MATERIALS AND METHODS
cording and reporting these types of lesions.
In the bioarcheology literature, periosteal new bone Pathological specimens from two London, U.K. pathol-
production is often mentioned as a result of specific in- ogy museums, St. George’s Hospital Pathology Museum
fectious disease, such as tuberculosis (e.g. Kelley et al., and the Hunterian Museum were selected for study. For
1994; Roberts et al., 1994), the treponematoses (e.g. inclusion in the research, specimens had to be long
Hackett, 1976; Schermer et al., 1994), and leprosy (e.g. bones exhibiting pathologically induced periosteal new
Lewis et al., 1995; Roberts and Manchester, 1997). Occa- bone, and had to be modern, pre-antibiotic, macerated,
sionally it is mentioned in association with syndromes dry, and unmounted.
such as hypertrophic osteoarthropathy (Fennell and St. George’s Hospital Pathology Museum is located
Trinkaus, 1997), or trauma (Detweiler-Blakely, 1988). within the Histopathology Department of the hospital’s
However, it is most frequently discussed as a product of medical school. The majority of the selected study speci-
nonspecific infection (e.g. Pfeiffer and Fairgrieve, 1994; mens had been in the museum’s collections prior to 1866
Larsen, 1997). (Ogle and Holmes, 1866) and were obtained from patient
When using the term ‘‘nonspecific infection’’ in this autopsy (see Table 3). The Hunterian Museum is located
way, it implies that an infection of some sort was visited within the Royal College of Surgeons of England and
upon an individual, its manifestation was a periosteal contains specimens collected by the surgeon and anato-
reaction, but the exact pathological organism involved mist John Hunter (1729–1793), those donated by other
in the infection is unknown. The ‘‘nonspecific infection’’ 18th and 19th century surgeons and physicians, and the
is thus not an entity unto itself; it is essentially an skeletons of executed felons acquired for anatomical
abbreviation of ‘‘I don’t know what caused this perios- demonstration (RCSE, 2006). Specimens examined from
teal new bone formation.’’ The majority of bioarcheology the Hunterian Museum are presented in Table 4.
papers offer few precise descriptions as to the location When employing pathology museum specimens in com-
or type of periosteal lesion present in an archaeological parative research, a number of factors must be taken
skeleton. Most papers simply state that ‘‘periostitis’’ into consideration. Pathology museum specimens usually
was present on the shaft of a tibia or another bone (e.g. consist of single bone elements and are displayed as
Lallo et al., 1978; Hodges, 1987; Palfi et al., 1992). The such. For example, if a researcher wanted to look at os-
terminology used by bioarcheologists to describe and seous syphilitic specimens, he/she would find skulls with
classify periosteal reactions has been inconsistent. A caries sicca displayed together in one case, single tibiae
number of researchers have devised classification sys- with syphilitic lesions displayed in another, and perhaps
tems for use in the analysis and description of perios- there may be a syphilitic fibula, ulna, or humerus. The
teal new bone production including: Lallo (1973); bones of complete individuals, if present, are rarely dis-
Strothers and Metress (1975); Cook (1976); Hackett played together, prohibiting the viewing of the overall
(1976); Lallo et al. (1978); Mensforth et al.(1978); Gra- distribution of skeletal lesions. There appears to be a
uer (1993), and Buikstra and Ubelaker (1994), but general lack of interest in the individual from whence
unfortunately no recording system has been universally the pathological specimen came. Age and sex data
adopted. regarding the specimen is rarely provided, nor is patient
Stage 1 Smooth and undamaged periosteal surface Represents appearance of tibia not yet afflicted
by infection
Stage 2 Longitudinal striations begin to appear in at least Initial stages of disease wherein periosteal damage and
three quarters*, two of which must be continuous destruction is only beginning
up and down
Stage 3 From 1 to 3 noncontinuous quarters with mild pitting
and swelling
Stage 4 At least 4 noncontinuous quarters with moderate
pitting and swelling
Stage 5 At least one local large swelling with moderate pitting From stage 5 onward periosteal damage and destruction
in three quarters becomes more acute
Stage 6 Three or more noncontinuous quarters exhibit swelling; or
two continuous quarters (up and down) exhibit swelling;
or one zone may exhibit swelling and scaling
Stage 7 Heavy pitting in at least one zone (zones may consist of
noncontinuous quarters)
Stage 8 Heavy pitting in at least two continuous quarters Stages 8 and 9 represent extremely heavy periosteal
destruction
Stage 9 Heavy pitting of periosteum over the entire area of three
continuous quarters and part of the fourth
* The bone was divided into four parts, each being defined as a ‘‘quarter’’: proximal epiphysis and metaphysis, proximal diaphysis,
distal diaphysis, distal metaphysis, and epiphysis.
TABLE 7. Types of periosteal reaction (Hackett, 1976) Packard Faxitron machine, while the Hunterian speci-
mens were radiographed with a Todd Research TR X-ray
Normal bone with surface changes
inspection cabinet machine. On both machines a tube
1. Normal bone with plaques
voltage of 50 kV was used, electrical current flow was
2. Normal bone with fine striation
3. Normal bone with superficial sequestra fixed at 3 mA, film to X-ray source distance was
4. Nodes/expansions with plaques 275 mm, and exposure times varied between 10 s and 1
5. Finely striate nodes/expansions min and 30 s, depending on the bone element and nature
6. Coarsely striate and pitted expansions of the pathology. The resulting radiographs were hand
7. Rugose nodes/expansions–slight changes, mostly nodes processed using the standard sequence of developer, stop
8. Rugose nodes/expansions–moderate changes, bath, fixer, and running water wash.
mostly expansions The radiographs were examined and the following in-
9. Rugose nodes/expansions–gross changes formation recorded: 1) appearance of the cortex (gross,
Nodes/Expansions with destruction permeated, geographic, or moth-eaten destruction;
10. Massive destruction–sequestra and expansions replacement of the cortex by periosteal new bone,
11. Focal destruction – nodes/expansions with superficial increased and decreased cortical thickness) (Greenfield,
cavitation
1986); 2) appearance of the endosteal surface (new bone
12. Focal destruction – metaphyseal expansion and cavitation
formation in the medulla and on the endosteal surface,
Expansions and deformity endosteal destruction, thickening of the endosteum, and
13. Platforms
bony bridges linking endosteal surfaces) (ibid.); 3) type
14. Bowing and expansion
of periosteal reaction (Edeiken et al., 1966; Edeiken,
1981) (see Table 8); 4) location of lesion(s); 5) size of
lesion(s) in mm; and 6) margination of lesion(s) (distinct
about the state of the underlying disease. Woven bone, or indistinct) (Greenfield, 1986). Most periosteal reac-
with its characteristic grey color, porous and disorgan- tions are easily recognized as opaque grey shadows
ized appearance, and sharp unremodeled edges, often occurring alongside the dense, white radiographic
appears to be resting on top of the cortical bone surface appearance of the cortical bone and often, but not
and is indicative of a disease process that was still active always, there is a distinct, thin black line that separates
at the time of death. Active lesions may have contributed the periosteal reaction from the cortex. All radiograph
to the death of an individual, but may not have neces- measurements were taken with sliding digital calipers.
sarily caused it. Lamellar bone, usually the same color To compensate for intra-observer error, all measure-
as the surrounding bone, with a more organized appear- ments were repeated three times with the average mea-
ance and rounded remodeled edges, indicates a healed surement recorded, and qualitative characteristics were
lesion, i.e. the person survived. A periosteal reaction con- recorded twice to see if consistency was maintained. The
sisting of both woven and lamellar bone indicates the majority of clinical papers regarding typologies of perios-
lesion was in the process of healing when the individual teal reaction are based on the radiographic analysis of
died (Roberts and Manchester, 1997). tumors (Edeiken et al., 1966; Edeiken, 1981; Ragsdale
et al., 1981; Ragsdale, 1993; Resnick, 1995), though the
Radiographic analysis authors state that their devised classifications are also
applicable to periosteal reactions associated with various
Each bone was radiographed using Kodak Industrex types of non-neoplastic diseases.
AA400 Ready Pack industrial X-ray film. The St. Statistical analysis of the macroscopic and radio-
George’s specimens were radiographed using a Hewlett- graphic results was conducted using one way analysis of
Fig. 1. Box plot comparing macroscopically visible lengths of Fig. 2. Box plot comparing macroscopically visible widths of
periosteal lesions. Transverse line represents median, box is periosteal lesions.
interquartile range, whiskers represent spread, circles represent
outliers, and stars represent extreme outliers.
RESULTS
Fifty-six bones were examined macroscopically and
radiographically (12 from the St. George’s Hospital Pa-
thology Museum and 44 from the Hunterian Museum)
and the macroscopic characteristics and roentgen fea-
tures of the periosteal lesions were recorded as outlined
above. The bones were grouped together under their re- Fig. 3. Box plot comparing macroscopically visible lesion
areas.
spective pathology museum disease categories: osteomye-
litis, necrosis, leg ulcer, syphilis, rickets, and ‘‘periostitis.’’
Macroscopic analysis focal lesions such as those seen in leg ulcers are usually
contained, smaller, localized reactions affecting single
The sizes of the periosteal lesions are of interest bone elements (Lippman and Goldin, 1961). As illus-
because the more diffuse a lesion is, the more likely it is trated by Figures 1 and 2, most of the ranges of both the
to result from a systemic pathological condition. Sys- lengths (parallel to the diaphysis) and widths (perpendic-
temic diseases such as syphilis, seem to have diffuse ular to the diaphysis) of the lesions in all disease catego-
lesions that are larger and more widely spread over the ries overlapped one another. Statistical analysis of the
surface of a bone (Chiu and Radolf, 1994). Conversely, macroscopic lesion length determined that there was a
* O 5 Osteomyelitis, N 5 Necrosis, U 5 Leg Ulcer, S 5 Syphilis, R 5 Rickets, P 5 ‘‘Periostitis’’. (Numbers were rounded to two dec-
imal points).
significant difference in lesion length between the dis- avoiding the proximal epiphysis. Periosteal lesions in
ease categories resulting from the greater lengths of the rickets overwhelmingly occurred in the midshaft area
syphilitic lesions when compared to those of the necrotic, (80%). Finally, periosteal lesions in ‘‘periostitis’’ were
ulcerous, and rachitic lesions (P \ 0.05). There were no seen primarily in the shaft (35%), although other areas
statistical differences seen among the lesion widths (P \ were also affected.
0.90, ANOVA). Periosteal lesion area was plotted (see Lesion types, whether consisting of woven bone, lamel-
Fig. 3), demonstrating that there were no statistically lar bone, or a mixture of the two, as previously stated
significant differences in the overall sizes of the perios- provide information about the unhealed/healing/healed
teal lesions between the disease categories (P \ 0.20, nature of the underlying disease. Lesion type did not
ANOVA). Aside from one outlier point, the necrotic bone show any specific patterning distinguishing the various
specimens have the most focal lesions. This relates well disease categories, as most of the lesions in the six dis-
to the clinical literature, which indicates that necrotic ease categories were of the mixed variety (see Table 10).
episodes in bone are usually localized events (Adler, This indicates that lesion type is related to the progres-
2000). Identifying the percentage area of the bone sive healing of the disease, rather than its nature. Simi-
affected, rather than the actual physical dimensions of larly, no patterns emerged with regard to the type of
the periosteal lesion, may better enable the comparison periosteal vascularization present among the lesions, as
of lesion size amongst different bones. most of the bones in all disease categories had lesions exhib-
The location of the periosteal lesion on the bone is im- iting mixed vascularization, a combination of large and
portant, as different diseases have characteristic lesion small striae and large and small foramina (see Table 10).
distributions (Ortner, 2003). Each bone specimen was di- The application of Lallo’s (1973) lesion severity stages
vided into seven sections, and the locations of the perios- was of little utility (see Tables 6 and 11) as most of the
teal lesions were identified within and across these lesions in all disease categories exhibited stage 9 sever-
sections (see Table 9). Osteomyelitis affected all of the ity. This was most likely due to the inherent bias in the
segments of the bones, with the entire bone being specimens themselves, as many had been selected for
affected the most frequently (21%). Necrosis most com- inclusion in the museums’ collections due to the severity
monly affected the distal third of the diaphysis (40%), of their lesions. The use of Hackett’s (1976) criteria for
among other areas. Leg ulcers were most commonly seen categorizing bone lesions (see Tables 7 and 12) yielded
affecting the distal metaphysis (29%), but other areas similar results, with none of the pathological conditions
were also affected. Syphilitic periosteal lesions were being characterized exclusively by one or multiple bony
most frequently seen affecting the bone shaft (63%) and traits.
TABLE 11. Frequency of lesion severity stages across disease because most of the radiographs were taken to illustrate
categories (after Lallo, 1973) the bones in the medio-lateral axis only.
Stages of lesion severity*
No relationship was seen between the different radio-
graphic categories of periosteal new bone production (as
1 2 3 4 5 6 7 8 9 described by Edeiken et al., 1966; Edeiken, 1981) and
O 0.00 0.00 0.08 0.00 0.08 0.00 0.00 0.23 0.62 specific pathological conditions (see Table 13), as most of
N 0.00 0.09 0.09 0.00 0.00 0.09 0.18 0.09 0.45 the disease categories had thin undulating periosteal
U 0.00 0.00 0.11 0.00 0.22 0.11 0.33 0.00 0.23 reactions. The margins of the periosteal lesions were
S 0.00 0.50 0.00 0.00 0.00 0.00 0.00 0.00 0.50 investigated as this indicates the rate of lesion growth
R 0.40 0.20 0.00 0.00 0.00 0.00 0.40 0.00 0.00 (Ragsdale et al., 1981), with radiographically indistinct
P 0.00 0.00 0.00 0.06 0.06 0.29 0.06 0.00 0.53
margins implying a faster growth rate (Greenfield,
* See Table 6 for descriptions of severity stages. 1986). Bones with osteomyelitis, leg ulcers, syphilis, and
rickets most often had periosteal lesions with indistinct
margins, while the margins of the necrotic bones’ perios-
Radiographic analysis teal lesions were most often distinct. The ‘‘periostitis’’
bones had almost equal incidences of distinct and indis-
The lengths and thicknesses of the periosteal lesions tinct periosteal lesions (see Table 14).
seen in the radiographs were plotted in Figures 4 and 5. The analysis of the state of the underlying bone corti-
Both the radiographic lengths and thicknesses of the ces indicated that there were no cortical characteristics
lesions in all disease categories appeared to overlap each that were specific to particular pathological conditions
other in range, but statistical analyses determined that (see Table 15). When examining the radiographic charac-
there were some significant differences across disease cat- teristics of the endosteal surfaces, it was difficult to
egories. For lesion length, differences resulted from the distinguish new bone formation from the effects of med-
smaller lengths of the rachitic lesions when compared to ullary destruction. Histological analysis of the endosteal
those of the ulcerous, and ‘‘periostitis’’ lesions (P \ 0.05). surface may help to distinguish between these two proc-
For lesion thickness, differences resulted from the smaller esses. The only distinctive endosteal feature involved the
thicknesses of the rachitic lesions when compared to those rachitic bone specimens, which had bridges of bone link-
of the osteomyelitic lesions (P \ 0.02). The radiological ing opposite endosteal surfaces. These bony bridges,
dimensions of periosteal lesions are associated with dis- occurring primarily in the diaphysis and in the area of
ease progression. This is especially true of lesion thick- the greatest bone curvature, appear to have been acting
ness, as the more chronic a lesion is, the thicker it will be as struts, supporting the bone in its unnaturally bent
(Dannels and Nashel, 1983; Resnick, 1995). state (see Table 16).
Similar to the macroscopic analysis, no patterns It was only possible to make direct comparisons of the
emerged with regard to the locations of the radiographi- macroscopic and radiographic results with regard to the
cally visible periosteal new bone production across the lengths and positions of the periosteal lesions. There was
various disease categories (see Table 9). The osteomyeli- a great deal of variation between the average lengths of
tis specimens were most often affected on the entire the periosteal lesions seen macroscopically and radio-
bone, from the proximal epiphysis to the mid-shaft, and graphically across the various disease categories (see
from metaphysis to metaphysis (all 19%), the necrotic Figures 1 and 4). Neither the average radiographic
specimens on the proximal metaphysis to the mid-shaft lengths nor the average macroscopic lengths were sys-
area (22%), the leg ulcer specimens from the midshaft to tematically greater - this was dependent on the disease
the distal metaphysis (28%), and the syphilitic speci- category. The syphilitic bones showed the most signifi-
mens from the proximal metaphysis to the distal meta- cant difference between macroscopic and radiographic
physis (33%). The rachitic specimens were affected lesion length (P \ 0.02), with the macroscopic lengths
equally at midshaft, the distal third of the diaphysis, being longer. Significant differences were also seen
from the proximal metaphysis to midshaft, from the among the ulcerous bones (P \ 0.05). The greatest
proximal third of the diaphysis to midshaft, and from amount of congruence occurred among the necrotic
the distal third of the diaphysis to the distal metaphysis bones, where the radiographic lengths were only slightly
(all 20%). Lastly, the ‘‘periostitis’’ specimens were most longer (P \ 0.75). No significant differences were seen
often affected on the bone shaft, from the proximal third among the rachitic (P \ 0.75), osteomyelitic (P \ 0.25),
of the diaphysis to the distal metaphysis, and from the or the bones labeled ‘‘periostitis’’ (P \ 0.10). One might
proximal third of the diaphysis to the distal epiphysis, expect the lesions to be longer on the radiographs, as it
and from the mid-shaft to the distal epiphysis (all 13%). is frequently difficult to discern the margins of the
It was not possible to investigate patterns of orientation lesions when analyzing the bones macroscopically, but
Fig. 4. Box plot comparing radiographically visible lengths Fig. 5. Box plot comparing radiographically visible thick-
of periosteal lesions. nesses of periosteal lesions.
surprisingly this was not always the case. The positions not suggest that periosteal new bone formation at any
of the periosteal lesions seen macroscopically or radio- location is uniquely diagnostic of any one pathological
graphically often differed, with the greatest degree of condition. Nor are the size, shape and form of periosteal
congruence occurring amongst the bones labeled ‘‘perios- new bone formation, as seen by the macroscopic exami-
titis’’ and the least amongst the ulcerous and necrotic nation of the bone surface or by radiography, uniquely
bones (see Table 9). However, when these differences diagnostic of any pathological condition. Many pathologi-
were tested statistically, no significant differences were cal conditions may cause periosteal new bone formation
seen (P \ 0.75 [O]; P \ 0.50 [N]; P \ 0.90 [U]; P \ 0.50 with no means of clearly diagnosing its etiology, and con-
[S]; P \ 0.75, P \ 0.10 [P]. versely, these same pathological conditions may be pres-
ent without periosteal new bone formation occurring.
DISCUSSION AND CONCLUSIONS The radiographic analyses of the pathology museum
bones emphasized the importance that radiography plays
So why was it so difficult to find macroscopic and radi- in the analysis of periosteal new bone. If the goal of an
ographic traits of periosteal new bone that were patho- analysis is to provide an accurate record of the amount
gnomonic of the pathological conditions seen in the periosteal new bone formation in a skeletal population,
pathology museum specimens? It appears that this lack practical implications aside, it is essential that all bones
of identifying traits is due to the way in which bone and be radiographed. As demonstrated by the comparison of
the periosteum responds pathologically. It is not the dis- macroscopically recorded and radiographically recorded
ease that determines how these tissues respond, it is the periosteal lesions lengths, there were great differences
tissues themselves. Pathological periosteal new bone is between the macroscopic and radiographic observations.
formed as a type of healing mechanism, whether as a Many periosteal lesions, particularly those that are
response to inflammation, mechanical adaptation/com- healed, are difficult to differentiate from normal compact
pensation due to osteolysis, tumor containment, or bone, and only become truly visible in radiographs. The
altered circulation (Ragsdale et al., 1981; Ragsdale, population prevalence of periosteal reactions may have
1993), and it appears that the agent that initiates the been underestimated in most archaeological skeletal pop-
periosteal reaction is largely irrelevant. Additionally, the ulations due to a lack of thorough radiographic investi-
presence of co-occurring diseases and an individual’s gation. It is unfortunate that systematic radiographic
age, sex, ethnicity, and immune and nutritional status analysis is prohibitively expensive, as this is compromis-
will influence how a disease manifests itself skeletally ing the acquisition of important data.
(Roberts and Manchester, 1997). The analysis highlighted the importance of analyzing
The clinical and bioarcheology literature and the path- complete skeletons when studying periosteal new bone
ological bones studied in the course of this research do formation in burial assemblages, as the examination of
TABLE 14. Frequency of distinct or indistinct periosteal lesion of specific infectious diseases, such as the treponemato-
margins across disease categories ses or leprosy.
No standardized, universally accepted terminology has
Distinct Indistinct
been devised to describe the various manifestations of
O 0.31 0.69 periosteal new bone production seen in archaeological
N 0.78 0.22 skeletons. Most bioarcheologists use a number of termi-
U 0.22 0.78 nologies, not clearly defined, which makes comparisons
S 0.00 1.00 between research results difficult. In published papers
R 0.00 1.00
P 0.56 0.44
and reports, precise descriptions of the location, or type,
of periosteal lesion present are usually lacking. Most
simply state that ‘‘periostitis’’ was present. The terminol-
ogy is extremely inconsistent between researchers, and
isolated bone elements may present a false impression of some use the qualifiers ‘‘mild,’’ ‘‘moderate,’’ and ‘‘severe’’
the nature of periosteal new bone formation throughout without defining what these terms mean. As previously
a skeleton. The varied results of the nature of the perios- mentioned, a number of researchers have attempted to
teal reactions seen in individual bones was illustrative of devise strategies for scoring periosteal reactions in skele-
this point and suggests that studies that derive preva- tal remains and all these methods have their strengths
lence rates for ‘‘nonspecific infections’’ based on the pres- and weaknesses, but most importantly they are all lack-
ence of periosteal reactions on individual bones may lead ing a consistency of use in the bioarcheology community.
to an overestimation of the prevalence of infectious dis- There has not been a method that has gained universal
ease in a population, as individual bones with periosteal acceptance as a recording standard, but Hackett’s (1976)
lesions could have been subject to trauma, localized system for classifying periosteal lesions as they relate to
ulceration, hypertrophic osteoarthropathy, or other non- the treponematoses appears to have been the most suc-
infectious conditions, for which differential diagnoses cessful (Elting and Starna, 1984; Reichs, 1989; Lewis,
may be ascertained with a complete skeleton. The perios- 1994; Schermer et al., 1994).
teal reactions seen in individual bones should not be It does not help that bioarcheological research involv-
diagnosed in isolation from the rest of the skeleton. ing periosteal reactions has been hampered by the lack
From the specimens described in this research, it is of interest in the topic by clinicians. This was a problem
clear that the term ‘‘periostitis’’ is fraught with difficul- as far back as 1817 when Crampton complained that
ties. In many practitioners’ minds, the term ‘‘periostitis,’’ ‘‘inflammation of the periosteum . . . [had not] been
or inflammation of the periosteum, is synonymous with noticed in any systemic work . . .’’ (Crampton, 1817, p.
the term ‘‘nonspecific infection.’’ This is patently untrue, 331). The trend has continued into the modern era, with
and may be derived from confusion between the terms medical texts still bereft of information. The majority of
‘‘inflammation’’ and ‘‘infection.’’ A bone that exhibits the clinical categorizations and descriptions of periosteal
periosteal new bone production did not necessarily come new bone formation are based almost entirely on the radi-
from an individual whose body was invaded by a patho- ological appearances of periosteal reactions associated
genic organism, nor does it even necessarily involve with neoplastic disease, a condition that is not a common
inflammation (Ragsdale, 1993; Ortner, 2003) as, to reit- occurrence in archaeologically derived skeletal material.
erate, periosteal new bone production can result from Clinicians rarely examine periosteal reactions as they
almost anything that breaks, tears, stretches, or even appear directly on bone, and pathologists often overlook
touches the periosteum (Richardson, 2001). The list of them during autopsy, unless specifically seeking them.
pathological conditions that can manifest periosteal new Periosteal new bone formation is generally considered a
bone production is long, but in spite of this, researchers sign of a disease—a sign that does not require any treat-
repeatedly equate ‘‘periostitis’’ with infection. ment. Moreover, because it does not require treatment, it
The problems associated with the equation of perios- is not worthy of significant study. This is unfortunate for
teal new bone formation with infection are compounded bioarcheologists because it leaves them with few clinical
by the coupling of ‘‘periostitis’’ with ‘‘nonspecific infec- models that can be applied to ancient skeletal material.
tion.’’ Powell (1988) stated the problem eloquently when In this respect, bioarchaeology has to generate its own
she wrote that: ‘‘. . . reporting the prevalence of infectious models for how periosteal reactions relate to various dis-
response or ‘‘periostitis’’ in general terms deliberately ease states, for it is only the bioarcheologists who truly
ignore[s] an aspect of population health of major signifi- see how periosteal reactions manifest themselves in dry
cance: the influence of specific infectious diseases upon bone specimens. For example, how are clinicians supposed
levels of morbidity and mortality.’’ By routinely recording to qualify and quantify periosteal reactions, when in their
periosteal new bone formation as ‘‘nonspecific infection,’’ initial manifestation, periosteal new bone formation is not
researchers could be potentially overlooking the effects visible radiologically (Shopfner, 1966).