How Does Tobacco Smoke Contribute to Cervical Carcinogenesis?

Tobacco smoke is a well-established human papillomavirus (HPV) cofactor for the development of cervical precancer and cancer (3, 6), but the molecular mechanisms by which smoking increases the risk of cervical precancer and cancer remain unknown. There are several plausible explanations (2). One is that smoking inhibits the immune response to HPV. A second is that carcinogenic HPV-infected cells are exposed to smoking carcinogens that cause DNA damage while HPV oncoproteins block apoptosis and cell cycle arrest. Alam et al. (1) reported a molecular interaction between benzo[a]pyrene (BaP), a carcinogen found in cigarette smoke, and HPV synthesis, suggestive of yet another possible mechanism. They found evidence that “high concentrations of BaP resulted in a ten-fold increase in HPV31 viral titers, whereas treatment with low concentrations of BaP resulted in increased HPV genome copies, but not virion morphogenesis”.

While Alam et al. (1) found that high concentrations induced high viral titers in their model system, it is notable that there was no evidence of a dose-response relationship between BaP concentration and HPV viral load that would lend biological plausibility. There is also no evidence that smokers have 1 μM BaP concentrations in their cervical tissue, which is more relevant than concentrations in cervical mucus.

Moreover, epidemiological studies have failed to consistently demonstrate an association of high HPV viral load, except for perhaps HPV type 16 (HPV16) (5), and incident cervical precancer or cancer. Nor has smoking been shown to elevate HPV viral load in humans. To examine the relationship of smoking and HPV viral load, data from atypical squamous cells of undetermined significance and a low-grade squamous-lesion triage study (ALTS)(8), a study in which cigarette smoking was shown to increase the risk of cervical precancer in carcinogenic HPV-positive women (6), were used. Using Hybrid Capture 2 signal strength as a semiquantitative measure of HPV viral load (4) among those women who were identified to have a single HPV genotype by PCR, no association of smoking status and semiquantitative HPV viral load was found for all women singly infected by any carcinogenic HPV genotype, by HPV31, or by HPV16 (Table 1). However, we note as a limitation that no method of viral load measurement that uses aliquots of exfoliated cervical cells can distinguish between two scenarios, lower viral load in many cells and higher viral load in fewer cells. It is plausible that smoking is associated with the latter condition and may be indicative of a higher risk of cervical precancer and cancer.

TABLE 1.

Comparison of the median Hybrid Capture 2 signal strengths, a semiquantitative measure of viral load (4), by smoking status and infection with a single carcinogenic HPV genotype in women participating in a low-grade squamous lesion triage study^a

Cytology	Smoking status	Women with any carcinogenic genotype				Women with HPV31				Women with HPV16
		Allb		<CIN2c		All		<CIN2		All		<CIN2
		n	RLU/PCd	n	RLU/PC	n	RLU/PC	n	RLU/PC	n	RLU/PC	n	RLU/PC
Alle	Never	842	155	664	171	74	159	49	145	146	132	85	74
	Former	149	148	107	239	17	96	14	94	37	71	20	92
	Current, <1 pack/day	408	155	281	151	35	49	19	44	106	135	44	85
	Current, 1-<2 packs/day	174	171	122	135	10	92	8	23	43	187	18	70
	Current, ≥2 packs/day	18	137	12	487	1	45	1	45	3	822	1	1,531
Negativef	Never	236	5	209	4	19	13	16	12	29	5	23	5
	Former	34	24	30	20	6	17	6	17	9	44	6	108
	Current, <1 pack/day	102	6	78	6	11	8	5	8	18	21	12	23
	Current, 1-<2 packs/day	47	8	42	6	4	16	3	9	7	10	6	6
	Current, ≥2 packs/day	3	38	3	38

^aA nonparametric test of trend was used to test for significance of smoking status and semiquantitative viral load.

^bAll women regardless of worst histology.

^c<CIN2, women with a worst 2-year histology result of less severe than cervical intraepithelial neoplasia grade 2.

^dHybrid Capture 2 signal strengths in relative light units (RLU) were compared with that of controls positive for 1 pg/ml HPV16 DNA (RLU/PC), and specimens with ≥1 RLU/PC were considered carcinogenic HPV DNA positive.

^eResults were nonsignificant at the following P values: women with any carcinogenic genotype, 0.9 (all) and 0.5 (<CIN2); women with HPV31, 0.3 (all) and 0.6 (<CIN2); women with HPV16, 0.1 (all) and 0.6 (<CIN2).

^fResults were nonsignificant at the following P values: women with any carcinogenic genotype, 0.6 (all) and 0.7 (<CIN2); women with HPV31, 0.2 (all) and 0.9 (<CIN2); women with HPV16, 0.8 (all) and 0.4 (<CIN2).

The HPV genome amplification due to exposure of HPV31b-infected raft cultures to low levels of BaP is an intriguing finding for the very reasons outlined by the authors (1), possible increased levels of the primary oncoproteins E6 and E7. Again, however, the authors did not find evidence of a dose-response effect, and only a twofold-higher effect than that for controls was observed. Unfortunately, there is a general lack of data on E6 and E7 expression and HPV natural history for reference. New assays that quantitatively measure HPV E6/E7 mRNA in human cervical specimens may be used to assess the relationship of smoking and E6/E7 expression.

To date, the molecular mechanism(s) by which smoking exerts an oncogenic effect on carcinogenic HPV-infected cells remains elusive. One recent study failed to find any difference in polycyclic aromatic hydrocarbon-DNA adduct formation between carcinogenic HPV-positive smokers and nonsmokers and between cases and controls (7). It is even uncertain at which stage(s) (HPV persistence, progression of persisting HPV infection to precancer, and/or invasion) smoking influences cervical carcinogenesis. Perhaps new tools, such as transgenic-mouse expression of HPV16 oncoproteins (9) or susceptible cell lines (10), may be useful for modeling the effects of smoking and smoking constituents on HPV biology. More laboratory and epidemiological research is needed to elucidate the etiologic role of smoking in cervical carcinogenesis.